Plant-based proteins as functional egg replacers in pound cake: A comparative study of legume and oat ingredients

TL;DR

This study compares plant-based proteins like legumes and oats as egg substitutes in pound cakes, finding that some ingredients offer promising but not fully egg-like performance.

Contribution

The paper provides a systematic comparison of fourteen plant-based proteins as egg replacers, highlighting their functional properties and cake quality outcomes.

Findings

Faba bean and lentil flours showed high protein solubility and good cake volume.

Oat flours provided strong emulsification stability and gelling capability.

No single plant protein fully replicated the multifunctionality of eggs in cake quality.

Abstract

Rising demand for plant-based egg alternatives is caused by the need for sustainable, ethical and allergen-free solutions to conventional egg use in food systems. A systematic comparison was conducted on fourteen plant protein ingredients, including faba bean, lentil, chickpea and oat protein isolates, concentrates and flours, as egg replacements in pound cake formulations. Key properties assessed included foaming capacity and stability, protein solubility, emulsification, sulfhydryl content, water- and oil-holding capacity and minimum gelling concentration, alongside cake quality metrics such as batter rheology, bake loss, specific volume and crumb texture. Results revealed marked differences among protein sources, with faba bean (FPF) and lentil (LPF) flours demonstrating high protein solubility (FPF: 83.46 ± 0.35; LPF: 79.56 ± 0.23%) and favourable specific volume (FPF: 2.13 ± 0.03; LPF: 2.18 ± 0.03 mL/g), chickpea protein concentrate (CPC) contributing desirable springiness (0.86 ± 0.03) and cohesiveness (0.60 ± 0.02) and oat flours (OPF, OPFF, OF) providing notable emulsification stability (OPF: 0.52 ± 0.07; OPFF: 0.53 ± 0.07; OF: 0.54 ± 0.05%/min) and gelling capability (OPF/OPFF/OF: 8%). While none of the plant protein ingredients fully matched the multifunctionality of eggs, several demonstrated encouraging textural properties within this pound cake model, highlighting promising potential for further optimisation. The findings emphasise the importance of ingredient selection, processing, and potential protein blending or modification to improve the performance of plant-based egg replacers, without yet achieving egg-like multifunctionality. Overall, this work refines practical understanding of how compositional and techno-functional differences among legume- and oat-based ingredients translate into pound cake quality, and outlines opportunities for future, more targeted formulation strategies.

Image 1

•Plant protein ingredients were systematically compared with whole egg powder.•Processing level strongly affected protein solubility, foaming and emulsification.•Less processed flours yielded cakes with better volume and texture.•Solubility and setting temperature correlated with cake specific volume.•Egg multifunctionality was not replicated by single plant ingredients.

Plant protein ingredients were systematically compared with whole egg powder.

Processing level strongly affected protein solubility, foaming and emulsification.

Less processed flours yielded cakes with better volume and texture.

Solubility and setting temperature correlated with cake specific volume.

Egg multifunctionality was not replicated by single plant ingredients.

Figures

Techno-functional properties.CEP = whole egg powder; FPI = faba bean protein isolate; FPC = faba bean protein concentrate; FPCD = faba bean protein concentrate deflavoured; FPF = faba bean protein flour; LPI = lentil protein isolate; LPC = lentil protein concentrate; LPF = lentil protein flour; CPC = chickpea protein concentrate; CPF = chickpea protein flour; CPFD = chickpea protein flour deflavoured; OPC = oat protein concentrate; OPF = oat protein flour; OPFF = oat protein flour fine; OF = whole grain oat protein flour.Values are presented as the average ± standard deviation, and bars with d

Physical properties of pound cake batters and pound cakes, and textural characteristics of pound cakes.CEP = whole egg powder; FPI = faba bean protein isolate; FPC = faba bean protein concentrate; FPCD = faba bean protein concentrate deflavoured; FPF = faba bean protein flour; LPI = lentil protein isolate; LPC = lentil protein concentrate; LPF = lentil protein flour; CPC = chickpea protein concentrate; CPF = chickpea protein flour; CPFD = chickpea protein flour deflavoured; OPC = oat protein concentrate; OPF = oat protein flour; OPFF = oat protein flour fine; OF = whole grain oat protein flour

Principal component analysis (PCA) plot of determined ingredient and respective cake attributes. Representing CEP = whole egg powder; FPI = faba bean protein isolate; FPC = faba bean protein concentrate; FPCD = faba bean protein concentrate deflavoured; FPF = faba bean protein flour; LPI = lentil protein isolate; LPC = lentil protein concentrate; LPF = lentil protein flour; CPC = chickpea protein concentrate; CPF = chickpea protein flour; CPFD = chickpea protein flour deflavoured; OPC = oat protein concentrate; OPF = oat protein flour; OPFF = oat protein flour fine; OF = whole grain oat protei

Full text

Introduction

1

The bakery sector faces a persistent challenge: identifying plant-based proteins that can replicate the structural and functional attributes of eggs in formulations such as pound cake. Eggs are uniquely effective in emulsifying, aerating, and delivering desirable crumb structure, rendering their replacement technically demanding and functionally complex (Halm et al., 2024, 2025). The rapid expansion of plant protein options, from purified isolates to minimally processed flours, offers new opportunities but introduces substantial variability in ingredient composition and performance. A systematic assessment of these alternatives is critical for guiding their optimal use in bakery applications.

Faba bean protein ingredients have attracted increasing interest as functional egg replacers in bakery formulations. Owing to their high protein content and balanced amino acid profile, faba bean concentrates have been utilised to enhance the nutritional and textural quality of cakes (Halm et al., 2025). Recent studies have shown that incorporating faba bean protein concentrates into pound cake can yield textural properties comparable to that of egg-based cakes, although processing steps such as deflavouring have been found to reduce functionalities such as protein solubility, foaming capacity, and oil-holding capacity (Halm et al., 2025). Hrabovska and Litvinov (2024) reported that the inclusion of faba bean concentrate in muffins improved the protein profile and nutritional value, lowered the glycaemic index, and produced texture and volume comparable to the control at moderate replacement levels. In another study, faba bean flour was used as a substitute for wheat flour in gluten-free cookies, resulting in bake loss and cookie volume similar to the control, along with acceptable sensory properties (Schmelter et al., 2021). Kumar et al. (2022) found that roasting raw faba bean flour enhanced water and oil absorption, emulsion stability, and gelling, thereby improving its functionality in bakery systems.

Lentil protein has emerged as another promising legume ingredient, offering a distinctive amino acid profile, suitable techno-functional properties, and widespread acceptance in plant-based bakery research (Jarpa-Parra et al., 2017; Shevkani et al., 2024). The replacement of egg protein with lentil protein in angel cake and muffin formulations did not significantly affect product volume or sensory scores, indicating that lentil protein is an effective foam and emulsion stabiliser in cake systems (Jarpa-Parra et al., 2017). Conversely, the inclusion of lentil protein can negatively impact dough properties, resulting in a firmer texture and darker crumb, while also presenting challenges associated with flavour and anti-nutritive compounds (Romano et al., 2021). Similarly, de la Hera et al. (de la Hera et al., 2012) observed that wheat-lentil composite flour in cakes reduced specific volume, cohesiveness, and springiness, while increasing hardness and lowering consumer acceptability.

Recent studies suggest that chickpea-derived ingredients can positively impact the textural, sensory, and nutritional properties of cakes. El Sohaimy et al. (El Sohaimy et al., 2021) enriched muffins with chickpea protein isolate, resulting in higher protein content and improved digestibility alongside high consumer acceptance, although texture analysis showed increased hardness, gumminess, and chewiness. In another study, chickpea aquafaba was used as a fat replacer in pound cake, increasing specific volume and resulting in no statistically significant differences in aroma, colour, texture, flavour or overall impression (Grossi et al., 2022). Similarly, chickpea aquafaba performed effectively as an egg substitute in cakes, demonstrating promising foam and emulsion properties, enhanced protein content, and favourable sensory attributes (Konal et al., 2025). Additionally, chickpea flour appears to be a promising ingredient in gluten-free baking due to its balanced functionality and health benefits (Vinod et al., 2023).

While legume-derived ingredients have demonstrated promising functionality in bakery applications, they often impart a beany flavour or cause colour changes that can limit sensory acceptance. To address these limitations, oat-based ingredients offer a complementary approach, providing balanced nutrition and a mild flavour profile. Research indicates that oat protein ingredients exhibit strong techno-functional properties, which can be enhanced through modifications such as enzymatic hydrolysis or chemical treatments (Holopainen-Mantila et al., 2024; Kumar et al., 2021). Sergiacomo et al. (2025) reported that incorporating 10–30% sprouted oat into sponge cakes influenced water absorption and emulsifying capacity, resulting in structural changes within the cake, while prolonged sprouting reduced specific volume but improved staling kinetics. In another study, varying levels of oat flour were used to replace wheat flour, leading to reductions in cake volume and modifications in texture, including decreased hardness, cohesiveness, and springiness. Despite these changes, consumers found the oat flour-containing cakes acceptable (De et al., 2013).

However, most previous studies have examined single plant sources or individual protein formats, typically using different recipes and processing conditions, which makes it challenging to clarify how processing level (isolate, concentrate, flour) and specific techno-functional properties together determine cake performance when eggs are replaced. There is still limited mechanistic understanding of how specific compositional traits and functional attributes of diverse plant ingredients translate into batter rheology and pound cake structure within a unified model system. Consequently, this limits formulation strategies for selecting and combining plant proteins to approximate egg functionality in cakes.

This study therefore tested the following hypotheses within a standardised pound cake model. First, processing level (protein isolate, concentrate, flour) and plant-protein source (faba bean, lentil, chickpea, oat) systematically affect protein solubility, foaming, emulsification, as well as water- and oil-holding and gelation behaviour. Second, key techno-functional properties of the ingredients are associated with cake performance indicators such as batter rheology (viscosity, batter changes during baking), specific volume and favourable crumb texture. Third, the plant protein ingredients that, based on their composition and techno-functional profile, most closely reproduce the behaviour of whole egg powder in this pound cake system can serve as promising bases for further optimisation (e.g., blending or enzymatic modifications) toward improved egg replacement strategies, even though full egg multifunctionality is not yet achieved. By testing these hypotheses, the work aims to move beyond a descriptive screening and provide mechanistic insight that can guide rational selection and processing of plant protein ingredients for egg replacement in pound cake.

Materials and methods

2

Materials

2.1

In this study, multiple plant protein sources were initially screened across several inclusion levels, and the protein level that yielded pound cakes with acceptable specific volume and crumb structure was selected as the optimal protein concentration for each ingredient (data not shown). The resulting optimal incorporation levels therefore varied among protein sources and were applied to the control recipe (Fig. 2). CEP was replaced with the respective plant protein ingredients at their determined optimal protein content for each source (Table 1). Because isolates, concentrates and flours inherently differ not only in protein content but also in starch, fibre and lipid levels, the cakes formulated with these ingredients also differed in overall matrix composition. As a result, the observed effects reflect the combined influence of protein functionality and non-protein components and cannot be attributed solely to protein in isolation. All protein ingredients were sourced from AGT Food and Ingredients (Regina, Saskatchewan, Canada), except for OPC, which was obtained from Lantmännen Biorefineries (Norrköping, Sweden). According to the manufacturers website, FPCD and CPFD were produced through a heat and moisture treatment. The control pound cake formulation included biscuit flour (Odlums Group, Dublin, Ireland), sucrose (Siucra, Dublin, Ireland), salt (Glacia, UK), sunflower oil (The King, Svitolovodsk, Ukraine), CEP (Igreca, Seiches sur le Loir, France), filtered tap water, and baking powder (Dr. Oetker, Bielefeld, Germany). CEP was replaced with the respective protein ingredients at their determined optimal protein content for each source (Table 1). Unless stated otherwise, all chemicals were obtained from Sigma-Aldrich (St. Louis, MO, USA).Table 1. Pound cake formulations.Table 1. Biscuit flourSucroseBaking powderSaltSunflower OilCEPProtein ingredientWaterCEPa10061.74.60.334.314.0-86.0FPI10061.74.60.334.3-9.390.7FPCa10061.74.60.334.3-12.987.1FPCDa10061.74.60.334.3-13.087.0FPF10061.74.60.334.3-24.076.0LPI10061.74.60.334.3-9.690.4LPC10061.74.60.334.3-14.086.0LPF10061.74.60.334.3-20.080.0CPC10061.74.60.334.3-11.188.9CPF10061.74.60.334.3-12.687.4CPFD10061.74.60.334.3-12.387.7OPC10061.74.60.334.3-12.787.3OPF10061.74.60.334.3-15.085.0OPFF10061.74.60.334.3-16.983.1OF10061.74.60.334.3-17.382.7CEP = whole egg powder; FPI = faba bean protein isolate; FPC = faba bean protein concentrate; FPCD = faba bean protein concentrate deflavoured; FPF = faba bean protein flour; LPI = lentil protein isolate; LPC = lentil protein concentrate; LPF = lentil protein flour; CPC = chickpea protein concentrate; CPF = chickpea protein flour; CPFD = chickpea protein flour deflavoured; OPC = oat protein concentrate; OPF = oat protein flour; OPFF = oat protein flour fine; OF = whole grain oat protein flour; All values are given based on flour (%).aHalm et al. (2025)

Protein ingredients

2.2

Compositional analysis

2.2.1

Compositional analysis was performed externally by Chelab S.r.l. (Resana, Italy) using standard reference methods. Moisture content was determined gravimetrically (AOAC Official Method 950.46B, Moisture in Meat, AOAC International, 1991). Protein content was measured by the combustion method (AOAC Official Method 992.23, Crude Protein in Cereal Grains and Oilseed Products, AOAC International, 17th Ed., 2000) using a nitrogen-to-protein conversion factor of 6.25 for all samples to ensure internal consistency and comparability with previous studies. Fat content was quantified by the Soxhlet method (AACC Method 30-25.01), and ash content was determined gravimetrically (AOAC 945.46). Sucrose content was measured by high-performance anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD) (ISO 22184 | IDF 244:2021). Total carbohydrate content was calculated by difference (AOAC 986.25) from measured moisture, protein, fat, ash, and sucrose contents, and is therefore a calculated value rather than a directly measured parameter. All results are expressed on a fresh weight (fw) basis.

Techno-functional analysis

2.2.2

Techno-functional properties were assessed in simplified model systems to characterise intrinsic behaviour of the ingredients relevant to pound cake manufacture. Foaming and emulsification tests evaluated air and oil-water interface stabilisation; water- and oil-holding capacity, protein solubility and minimum gelling concentration related to batter viscosity, water distribution and network formation; sulfhydryl (SH) group content was used as an indicator of potential for disulfide bond formation and protein network development; and pH and total titratable acidity (TTA) described the acid-base environment that can influence protein charge, solubility and leavening reactions.

Foam capacity and stability

2.2.2.1

Foam capacity and stability were determined according to the method of Halm et al. (2025). Dispersions (2% w/w) were prepared in distilled water, and the pH was adjusted to 7.0 using HCl or NaOH. Samples were hydrated overnight at 4 °C, then brought to room temperature. Sediment was carefully redispersed, and pH was rechecked and readjusted if necessary. The initial sample height was recorded prior to foaming. Foaming was performed using an Ultra-Turrax with an S10N-10G dispersing element (IKA Labortechnik, Janke & Kunkel GmbH, Staufen, Germany) at maximum speed for 30 s. Foam height was measured immediately (0 min) and after 1 h. Foam capacity was expressed as the percentage increase in volume immediately after mixing, and foam stability as the percentage of foam remaining after 1 h, relative to the initial foam height. The following equations were applied:

$$[eqn]$$ $$[eqn]$$

Protein solubility

2.2.2.2

Protein solubility was determined according to the method of Jaeger et al. (2023). Protein dispersions (1% w/w) were prepared in distilled water, and the pH was adjusted to 7.0 using HCl or NaOH. Samples were hydrated overnight at 4 °C with continuous shaking. The pH was then rechecked and readjusted to 7 if necessary, followed by centrifugation at 4000 rpm for 30 min. The protein content of the supernatants was quantified by the Kjeldahl method (N × 6.25). Protein solubility was expressed as the percentage of protein present in the supernatant.

Emulsifying characteristics

2.2.2.3

Emulsifying properties were determined according to the method of Halm et al. (2025). Aqueous dispersions containing 1% protein (w/w) were prepared in distilled water, and the pH was adjusted to 7 using HCl or NaOH. Samples were hydrated overnight at 4 °C with continuous shaking. Emulsions were prepared by mixing the protein dispersions with sunflower oil at a 90:10 ratio (v/v) and homogenizing with an Ultra-Turrax equipped with an S10N-10G dispersing element (IKA Labortechnik, Janke & Kunkel GmbH, Staufen, Germany) at maximum speed (setting 5) for 2 min. Emulsion stability was evaluated using an analytical centrifuge (LUMiSizer, LUM GmbH, Berlin, Germany) at 100 rcf and 15 °C. Results were expressed as separation rate (%/min) and transmission profiles over the full measurement range.

Sulfhydryl (SH) groups

2.2.2.4

The concentration of sulfhydryl (SH) groups in the protein ingredients was determined using a method described by Halm et al. (2025). Ellman's reagent [5,5′-dithiobis(2-nitrobenzoic acid); DTNB], which reacts at neutral or alkaline pH with protein SH groups to form thionitrophenylated protein and a yellow thionitrophenylate anion was used for total SH group determination. 40 μL of Ellman's reagent was added to 4 mL of the protein dispersion (sample) or Tris–Gly buffer (pH 8, 8 M urea) (blank). After 15 min incubation at ambient temperature, the absorbance of the supernatant was measured at 412 nm. SH group concentration was expressed as μmol SH per gram of protein, calculated according to:

$$[eqn]$$

Where ABS_sample_ is the absorbance of the sample at 412 nm, ABS_blank_ is the absorbance of the blank at 412 nm, ABS_reagent blank_ is the absorbance of the reagent blank at 412 nm, ε is the extinction coefficient (13,600 M^−1^cm^−1^), and C is the protein concentration in mg/mL of the diluted sample.

pH and total titratable acidity (TTA)

2.2.2.5

pH and total titratable acidity (TTA) were determined according to the method described by Halm et al. (2025). 10 g of sample was mixed with 95 mL of distilled water and 5 mL of acetone, then stirred thoroughly. pH was measured using a pH meter (Mettler Toledo, Columbus, OH, USA). Samples were titrated with 0.1 M NaOH to an endpoint of pH 8.5. After a 3 min pause, the pH was rechecked and adjusted if necessary. TTA was expressed as millilitres of 0.1 M NaOH required to reach the endpoint, per 10 g of sample.

Water- and oil-holding capacity

2.2.2.6

Water-holding capacity (WHC) and oil-holding capacity (OHC) were determined according to the method of Halm et al. (2025). 1 g of sample was weighed into pre-weighed tubes, followed by the addition of 6 g of distilled water (for WHC) or sunflower oil (for OHC). Samples were vortexed for 3 min, incubated at room temperature for 1 h, and centrifuged at 4000×g for 30 min at 20 °C. The supernatant was discarded, and the tubes were inverted onto a paper towel for 30 min to drain excess liquid. The final weight was recorded, and WHC and OHC were calculated according to the following equation:

$$[eqn]$$

Minimum gelling concentration

2.2.2.7

Minimum gelling concentration was determined according to the procedure of Vogelsang-O’Dwyer et al. (Vogelsang-O’Dwyer et al., 2020) with minor modifications. Protein dispersions (6-30% w/w) were prepared in 2% increments (6, 8, 10, …, 30%). The appropriate amount of sample was weighed into tubes, and the total mass was adjusted to 15 g with distilled water. The pH was adjusted to 7.0 using HCl or NaOH solutions ranging from 0.01 M to 2 M, and samples were hydrated overnight at 4 °C. Dispersions were then heated in a water bath at 90 °C for 30 min, rapidly cooled on ice for 10 min, and incubated overnight at 4 °C. Samples were then inverted, and the lowest protein concentration at which the dispersion did not flow within 30 s was recorded as the minimum gelling concentration.

Cake analysis

2.3

Cake preparation and analysis were performed according to Halm et al. (2025) with some modifications. In this study, cake quality was evaluated using instrumental measurements of cake batter, specific volume and texture profile analysis. No formal sensory or consumer evaluation was conducted.

Cake preparation

2.3.1

The formulations of the different cakes are presented in Table 1. For the control cake, water was added to CEP and premixed using a Kenwood Chef (Kenwood, Havant, UK) fitted with a whisk attachment (60 s, minimum speed). For the plant-based protein pound cakes, dispersions of the plant protein ingredient in water were prepared using an Ultra-Turrax (IKA-Werke GmbH & Co. KG, Staufen, Germany) at 12,000 rpm for 5 min. Dry ingredients (flour, baking powder, sucrose, and salt) were combined, after which the liquid ingredients (egg mixture or plant protein dispersion, sunflower oil) were added and mixed (120 s, minimum speed). This mixing protocol and equipment were kept constant for all formulations to ensure comparable mixing energy and air incorporation across the egg and plant-protein batters. A 250 g portion of cake batter was transferred to a greased baking pan (height: 58.14 mm; top length: 149.62 mm; bottom length: 138.46 mm; top width: 85.16 mm; bottom width: 75.22 mm). The batters were baked in a preheated deck oven (MIWE condo, Arnstein, Germany) at 180 °C for 55 min. After baking, cakes were cooled at room temperature for 2 h before further analysis.

Batter changes during baking: micro-baking

2.3.2

Changes in the viscoelastic properties of the cake batter during heating were assessed using the micro-baking test described by Schirmer et al. (2012). The rheometer (Physica MCR 301, Anton Paar Group, Graz, Austria) was equipped with a plate-plate geometry (50 mm diameter) and a fixed gap of 2 mm. The batter sample was carefully applied to the lower plate, and after the measuring gap was adjusted, excess material at the edges was trimmed to obtain a uniform sample geometry. Measurements were conducted at a strain of 0.1% and a frequency of 1 Hz while the temperature was increased from 25 to 98 °C at 2.4 °C/min. The complex modulus (G∗) was recorded as a function of temperature, and the inflection temperature (T_i_) and storage modulus at T_i_ (G'i) were used as indicators of structure-setting behaviour.

Cake batter viscosity

2.3.3

The viscosity of the cake batter was measured using a Physica MCR 301 rheometer (Anton Paar Group AG, Graz, Austria), as previously described by Halm et al. (2025). Batter samples were transferred into the concentric cylinder (CC) measurement system and filled to the indicated level. A measurement probe with a diameter of 26.7 mm was used, and the apparent viscosity was recorded at a shear rate of 10 s^−1^ and a temperature of 20 °C.

Bake loss

2.3.4

The bake loss (%) during baking was determined by weight difference before and after baking.

Specific volume

2.3.5

The specific volume was determined using a VolScan Profiler (Stable Micro Systems, Godalming, UK).

Textural analysis

2.3.6

Crumb texture was evaluated using texture profile analysis (TPA) with a TA-XT2i texture analyser (Stable Micro Systems, Godalming, UK) equipped with a flat-ended cylindrical aluminium probe (50 mm diameter) and a TA-90 platform, as described by Halm et al. (2025). Cakes were sliced into 25 mm-thick sections, excluding the end pieces, yielding four slices per cake for analysis. One measurement was performed per slice, resulting in four measurements per cake. For each formulation, two cakes were prepared per batch, and three independent batches were produced, giving a total of 24 measurements. A two-cycle compression test was conducted with a pre-test speed of 1.0 mm/s, a test speed of 5.0 mm/s, a post-test speed of 10 mm/s, and a compression distance of 10 mm.

Statistical analysis

2.4

Each formulation was prepared in three independent batches (n = 3), with two cakes baked per batch. Unless otherwise stated, all cake measurements were performed on both cakes from each batch, and the reported values represent the mean of the three independent batches. For data exhibiting a normal distribution, one-way analysis of variance (ANOVA) was carried out (p ≤ 0.05), followed by Tukey's post hoc test using OriginPro© 2025 (Version 10.2.0.196, Learning Edition; OriginLab Corporation, Northampton, MA, USA). When the data did not meet normality assumptions, the nonparametric Kruskal–Wallis test was applied (p < 0.05). Principal Component Analysis (PCA) was performed using OriginPro to identify correlations among variables and reduce data dimensionality. Regression and Pearson correlation analyses were conducted in Microsoft Excel (2021) (Microsoft Corporation, Redmond, WA, USA) to examine relationships between the functional properties of the ingredients and the cake analysis results.

Results

3

Protein ingredients

3.1

Compositional analysis

3.1.1

The compositional analysis revealed distinct differences among all sample types (Table 2). CEP (control egg powder) contained 5.07 ± 0.46 g/100g moisture, 53.40 ± 2.10 g/100g protein, 34.20 ± 2.00 g/100g fat, 3.76 ± 0.25 g/100g ash, 3.57 ± 2.95 g/100g carbohydrates (by difference), which represent the remaining solids fraction after moisture, protein, fat, ash and sucrose are accounted for, and no detected sucrose, as previously reported by Halm et al. (2025). Faba bean-based ingredients demonstrated varying compositions: FPI (faba bean protein isolate) showed high protein (80.60 ± 0.37 g/100g) and moisture (8.60 ± 0.70 g/100g), low fat (7.14 ± 0.84 g/100g), and no detected carbohydrates or sucrose. FPC and FPCD (faba bean protein concentrate and its deflavoured variant) had similar protein (58.00 ± 2.30 g/100g and 57.90 ± 2.30 g/100g, respectively) and fat (4.06 ± 0.48 g/100g and 2.97 ± 0.35 g/100g) contents (Halm et al., 2025), while FPF (faba bean protein flour) exhibited a lower protein content (31.30 ± 1.40 g/100g) and a higher carbohydrate content (55.46 ± 1.58 g/100g).Table 2. Compositional analysis.Table 2[g/100g]MoistureProteinFatAshCarbohydrates (by difference)SucroseCEPa5.07 ± 0.4653.40 ± 2.1034.20 ± 2.003.76 ± 0.253.57 ± 2.95n.d.FPI8.60 ± 0.7080.60 ± 0.377.14 ± 0.845.30 ± 0.36n.d.n.d.FPCa6.28 ± 0.5158.00 ± 2.304.06 ± 0.485.81 ± 0.3925.85 ± 2.441.82 ± 0.18FPCDa6.35 ± 0.5257.90 ± 2.302.97 ± 0.355.66 ± 0.3827.12 ± 2.411.62 ± 0.16FPF8.27 ± 0.6731.30 ± 1.401.62 ± 0.193.35 ± 0.2355.46 ± 1.582.01 ± 0.20LPI5.93 ± 0.4978.30 ± 3.006.51 ± 0.769.19 ± 0.62n.d.n.d.LPC6.84 ± 0.5653.70 ± 2.103.86 ± 0.455.28 ± 0.3530.32 ± 2.251.80 ± 0.18LPF8.27 ± 0.6725.00 ± 1.101.38 ± 0.162.61 ± 0.1862.74 ± 1.311.76 ± 0.17CPC3.77 ± 0.3167.30 ± 2.6022.30 ± 1.406.34 ± 0.430.29 ± 3.290.04 ± 0.01CPF8.78 ± 0.7219.88 ± 0.956.67 ± 0.782.52 ± 0.1762.15 ± 1.433.09 ± 0.30CPFD5.57 ± 0.4620.33 ± 0.965.74 ± 0.962.42 ± 0.1665.94 ± 1.273.09 ± 0.30OPC3.43 ± 0.3258.90 ± 2.3015.90 ± 1.304.11 ± 0.2817.66 ± 2.680.66 ± 0.06OPF9.63 ± 0.3516.72 ± 0.862.93 ± 0.341.97 ± 0.1368.75 ± 1.001.20 ± 0.12OPFF9.07 ± 0.3314.82 ± 0.812.07 ± 0.241.71 ± 0.1272.33 ± 0.911.16 ± 0.11OF10.88 ± 0.3014.45 ± 0.806.81 ± 0.801.92 ± 0.1365.94 ± 1.181.09 ± 0.11CEP = whole egg powder; FPI = faba bean protein isolate; FPC = faba bean protein concentrate; FPCD = faba bean protein concentrate deflavoured; FPF = faba bean protein flour; LPI = lentil protein isolate; LPC = lentil protein concentrate; LPF = lentil protein flour; CPC = chickpea protein concentrate; CPF = chickpea protein flour; CPFD = chickpea protein flour deflavoured; OPC = oat protein concentrate; OPF = oat protein flour; OPFF = oat protein flour fine; OF = whole grain oat protein flour.Carbohydrate values are calculated by difference from measured moisture, protein, fat, ash, and sucrose contents and therefore represent the remaining solids fraction (including starch and unquantified fibre).n.d. = not detected.aHalm et al. (2025)

Lentil-based protein ingredients also varied: LPI (lentil protein isolate) and LPC (concentrate) contained 78.30 ± 3.00 g/100g and 53.70 ± 2.10 g/100g protein, respectively, with low fat levels and no detected carbohydrates or sucrose in LPI. LPF (lentil protein flour) had 25.00 ± 1.10 g/100g protein and a carbohydrate content of 62.74 ± 1.31 g/100g. Among the chickpea-derived ingredients, CPC (chickpea protein concentrate) contained 67.30 ± 2.60 protein and 22.30 ± 1.40 g/100g fat, while CPF (chickpea protein flour) and CPFD (its deflavoured version) showed lower protein levels (19.88 ± 0.95 g/100g and 20.33 ± 0.96 g/100g, respectively), and carbohydrate values of 62.15 ± 1.43 g/100g and 65.94 ± 1.27 g/100g, respectively.

For the oat-based products, OPC (oat protein concentrate) had 58.90 ± 2.30 g/100g protein and 15.90 ± 1.30 g/100g fat, with 17.66 ± 2.68 g/100g carbohydrates. OPF (oat protein flour), OPFF (fine flour), and OF (whole grain oat protein flour) displayed decreased protein contents (16.72 ± 0.86 g/100g, 14.82 ± 0.81 g/100g, and 14.45 ± 0.80 g/100g), while carbohydrate content was highest in OPFF (72.33 ± 0.91 g/100g) and OF (65.94 ± 1.18 g/100g). Sucrose was not detected in several isolates but present at low levels across most flours and concentrates.

Techno-functional analysis

3.1.2

Foam capacity and stability

3.1.2.1

Foam capacity values (Fig. 1, Table S-1) showed that LPI (48.00 ± 0.00%) was significantly higher than all other samples. FPC (39.74 ± 0.45%) also exhibited a foam capacity significantly greater than most of the other ingredients. CEP (26.14 ± 5.06%) and FPI (24.99 ± 2.22%) displayed intermediate foam capacities, with no significant difference between them. FPCD (18.67 ± 1.15%), LPC (20.00 ± 0.00%), LPF (17.33 ± 2.31%), and FPF (16.00 ± 0.00%) all showed comparable foam capacities, without significant differences among their values. Lower foam capacities were observed for CPC (10.67 ± 1.15%), CPF (10.20 ± 0.00%), OPF (9.44 ± 0.96%), OPFF (10.00 ± 0.00%), and OF (4.17 ± 0.00%), with no significant differences detected within this lowest range.Fig. 1. Techno-functional properties.CEP = whole egg powder; FPI = faba bean protein isolate; FPC = faba bean protein concentrate; FPCD = faba bean protein concentrate deflavoured; FPF = faba bean protein flour; LPI = lentil protein isolate; LPC = lentil protein concentrate; LPF = lentil protein flour; CPC = chickpea protein concentrate; CPF = chickpea protein flour; CPFD = chickpea protein flour deflavoured; OPC = oat protein concentrate; OPF = oat protein flour; OPFF = oat protein flour fine; OF = whole grain oat protein flour.Values are presented as the average ± standard deviation, and bars with different lowercase letters within each panel differ significantly (P < 0.05).Fig. 1. Fig. 2Appearance and average specific volume of the pound cakes containing CEP = whole egg powder; FPI = faba bean protein isolate; FPC = faba bean protein concentrate; FPCD = faba bean protein concentrate deflavoured; FPF = faba bean protein flour; LPI = lentil protein isolate; LPC = lentil protein concentrate; LPF = lentil protein flour; CPC = chickpea protein concentrate; CPF = chickpea protein flour; CPFD = chickpea protein flour deflavoured; OPC = oat protein concentrate; OPF = oat protein flour; OPFF = oat protein flour fine; OF = whole grain oat protein flour.Values below each image represent the protein content (% w/w) in the aqueous dispersions of the respective plant protein ingredients used in the formulations. Specific volumes (mL/g) of the cakes are indicated above the images.Fig. 2

Foam stability results demonstrated that LPC and LPF (each 100.00 ± 0.00%) reached the highest values, which were significantly higher than most other samples. FPF (95.83 ± 7.22%) and FPCD (82.22 ± 5.88%) were also among the samples with the highest stabilities. CEP (74.92 ± 4.50%), FPC (71.67 ± 2.89%), CPC (75.56 ± 7.70%), and CPF (80.00 ± 0.00%) exhibited moderately high foam stability, showing significant differences mainly when compared to samples with lower stabilities. FPI (62.24 ± 7.79%), CPFD (62.04 ± 11.23%), OPC (37.50 ± 0.00%), and LPI (29.17 ± 0.00%) had lower foam stabilities, with OF (0.00 ± 0.00%) being significantly lower than all other samples. No significant differences were observed within the lower stability group.

Given that aeration and gas-cell stabilisation during mixing and early baking are critical for pound cake volume, these foaming data provide an indication of the intrinsic ability of the ingredients to support air incorporation, although interactions with wheat flour and other batter components are not captured in this system (Halm et al., 2024, 2025).

Protein solubility

3.1.2.2

Protein solubility at pH 7 varied considerably among the samples (Fig. 1, Table S-1). CEP exhibited the highest solubility (87.75 ± 1.15%), which was significantly higher than all other ingredients. FPF (83.46 ± 0.35%) demonstrated the next highest value, followed by FPC (78.37 ± 0.47%) and LPF (79.56 ± 0.23%), with no significant difference between FPC and LPF. Moderate solubility was observed in FPCD (51.88 ± 0.32%), LPC (47.16 ± 0.50%), CPF (56.21 ± 0.63%), and CPFD (55.40 ± 0.49%), where no significant difference was detected between CPF and CPFD. Significantly lower protein solubility was found in FPI (6.18 ± 0.29%), OPF (5.89 ± 0.14%), OPFF (5.45 ± 0.15%), OF (6.09 ± 0.03%), CPC (8.34 ± 0.07%), and LPI (8.91 ± 0.26%), which showed no significant differences among themselves. The lowest solubility was recorded for OPC (2.91 ± 0.17%), significantly lower than all other samples.

Emulsifying characteristics

3.1.2.3

The emulsifying characteristics, assessed by separation rate, is displayed in Fig. 1 and Table S-1. CEP had the lowest separation rate (0.17 ± 0.03%/min). OPF (0.52 ± 0.07%/min), OPFF (0.53 ± 0.07%/min), and OF (0.54 ± 0.05%/min) also exhibited low separation rates, indicating good emulsion stability characteristics, CPF (0.96 ± 0.02%/min), LPF (1.19 ± 0.09%/min), CPFD (1.13 ± 0.21%/min), CPC (1.30 ± 0.04%/min), and FPF (1.39 ± 0.27%/min) presented moderate separation rates. Higher separation rates were observed for LPC (1.87 ± 0.09%/min), OPC (1.93 ± 0.06%/min), FPI (1.96 ± 0.04%/min), LPI (2.12 ± 0.06%/min), and FPCD (2.12 ± 0.10%/min), which were significantly higher than most other samples. FPC (1.55 ± 0.21%/min) showed a separation rate that was intermediate, and not significantly different compared to the higher and moderate groups.

Because fat distribution and oil-water interface stability can influence gas-cell stability and crumb structure, these emulsion stability measurements are used as a proxy for how different ingredients may affect fat structuring in cake, recognising that the simplified oil-water system differs from the full batter mix (Halm et al., 2024; Karaca et al., 2011).

Sulfhydryl (SH) groups

3.1.2.4

Sulfhydryl group (SH-group) content differed significantly among the samples (Fig. 1, Table S-1). CEP exhibited the highest value (196.03 ± 2.29 μmol SH/g), significantly surpassing all other ingredients. Oat-derived ingredients (OPC, OPF, OPFF, OF) followed, with high and statistically similar SH-group contents (133.39 - 143.03 μmol SH/g), all of which were still significantly lower than CEP. CPF (95.72 ± 2.63 μmol SH/g) and CPFD (84.62 ± 6.87 μmol SH/g) showed moderate values, differing significantly from both the oat group and the majority of legume-based samples. The lowest SH-group contents were observed in most legume ingredients, ranging from 51.66 ± 3.44 to 72.20 ± 4.55 μmol SH/g, with no significant differences detected among them.

Because disulfide bond formation contributes to protein network strength and interfacial film stability, SH group content provides complementary information on the potential of each ingredient to support gel and network formation in cakes (Wu et al., 2021; De Angelis et al., 2024a).

pH and total titratable acidity (TTA)

3.1.2.5

The values for pH and total titratable acidity (TTA) are depicted in Fig. 1 and Table S-1. Across most samples, pH values were very similar, ranging from 6.18 ± 0.01 to 6.77 ± 0.00, with no significant differences detected between the plant-based ingredients. In contrast, CEP had a significantly higher pH (8.68 ± 0.01) than all other samples, while TTA was not detectable. Among the plant-based ingredients, the highest TTA was observed in FPI (29.90 ± 0.10 mL/10g), followed by FPC (28.40 ± 0.49 mL/10g), LPI (24.70 ± 0.10 mL/10g), CPC (23.80 ± 0.13 mL/10g), LPC (21.63 ± 0.26 mL/10g), and FPCD (19.28 ± 0.25 mL/10g). Lower TTA values were found in FPF (12.50 ± 0.13 mL/10g), OPC (11.26 ± 0.30 mL/10g), LPF (10.48 ± 0.12 mL/10g), CPF (8.83 ± 0.03 mL/10g), CPFD (7.33 ± 0.08 mL/10g), and the lowest in OPF (4.25 ± 0.30 mL/10g), OPFF (3.95 ± 0.23 mL/10g), and OF (4.32 ± 0.03 mL/10g), with significant differences throughout this lower range.

pH and TTA are important because they determine protein charge state, solubility and interactions with leavening agents; shifts in the acid-base balance can thus affect gas release, batter viscosity and protein–starch network development during baking (Halm et al., 2024, 2025).

Water- and oil-holding capacity

3.1.2.6

The water-holding capacity (WHC) of the samples exhibited significant differences, ranging from 53.22 ± 0.53% for FPC to 280.06 ± 2.14% for FPI (Fig. 1, Table S-1). LPI also showed a high water-holding capacity at 274.33 ± 0.58%. Lower values were observed for samples such as CPF (73.65 ± 2.05%) and FPF (63.91 ± 0.83%). Intermediate water-holding capacity values were recorded for the other sample groups, with significant differences observed among most of the samples. Oil-holding capacities (OHC) also varied significantly among the samples, with values spanning from 67.59 ± 4.26% for LPF to 104.65 ± 1.05% for FPC (Table S-1). Samples such as LPF (67.59 ± 4.26%) and FPF (67.52 ± 1.43%) demonstrated lower oil-holding capacities, while OF and OPC presented higher values of 87.53 ± 0.61% and 93.38 ± 4.74%, respectively. WHC and OHC are relevant for batter viscosity and moisture retention during baking (Halm et al., 2025; Buhl et al., 2019).

Minimum gelling concentration

3.1.2.7

The minimum gelling concentrations of the samples, which is related to network formation during heating (Halm et al., 2025; Sun et al., 2023), varied notably (Fig. 1, Table S-1). OF, OPF, and OPFF all exhibited the lowest minimum gelling concentration at 8%. FPC showed a relatively low value of 12%, while CEP, LPC, CPFD each demonstrated concentrations of 10% or 14%. In contrast, higher minimum gelling concentrations were observed for FPCD and LPI at 18%, and for FPI and FPF at 16%. The highest concentrations were recorded for LPF, CPC, and CPF, ranging from 24% to 26%.

Cake analysis

3.2

Batter changes during baking: micro-baking

3.2.1

The temperature at the inflection point (T_i_), indicating the transition of the cake batter from liquid to solid state, ranged from 77.03 ± 0.65 °C to 80.50 ± 0.66 °C (Fig. 3, Table S-2). FPF and LPF exhibited the highest T_i_ values (80.50 ± 0.66 °C), whereas OPC had the lowest (77.03 ± 0.65 °C). Several samples, including FPI, FPCD, CPF, CPFD, OPF, OPFF, and OF, had similar T_i_ values, all between 77.42 ± 0.04 °C and 77.90 ± 0.62 °C, indicating no significant differences among these groups. LPI, LPC, and CPC showed intermediate T_i_ values from 78.28 ± 0.65 °C to 78.90 ± 0.49 °C, again with no significant differences detected.Fig. 3. Physical properties of pound cake batters and pound cakes, and textural characteristics of pound cakes.CEP = whole egg powder; FPI = faba bean protein isolate; FPC = faba bean protein concentrate; FPCD = faba bean protein concentrate deflavoured; FPF = faba bean protein flour; LPI = lentil protein isolate; LPC = lentil protein concentrate; LPF = lentil protein flour; CPC = chickpea protein concentrate; CPF = chickpea protein flour; CPFD = chickpea protein flour deflavoured; OPC = oat protein concentrate; OPF = oat protein flour; OPFF = oat protein flour fine; OF = whole grain oat protein flour.Values are presented as the average ± standard deviation, and bars with different lowercase letters within each panel differ significantly (P < 0.05).Fig. 3

For the storage modulus (G'i), results ranged from 258.00 ± 29.15 Pa to 821.17 ± 127.45 Pa. The highest G'i values were observed for FPF (747.17 ± 141.00 Pa) and LPF (821.17 ± 127.45 Pa). FPC and OF also showed relatively high moduli (575.50 ± 42.12 Pa and 503.00 ± 28.26 Pa), while a large group consisting of FPI, FPCD, LPI, LPC, CPC, CPF, CPFD, OPC, OPF, and OPFF had G'i values ranging from 359.33 ± 80.13 Pa to 439.00 ± 17.06 Pa, with no significant differences within this group. The lowest G'i was recorded for CEP (258.00 ± 29.15 Pa).

Cake batter viscosity

3.2.2

The apparent viscosity of the cake batters demonstrated a wide range of values across the different samples (Fig. 3, Table S-2). The highest viscosity was observed for LPF (35.83 ± 3.05 Pa s). OF and CPFD also exhibited relatively high viscosities at 24.12 ± 1.36 Pa s and 24.41 ± 1.01 Pa s, respectively, with no significant difference between these two samples. FPF displayed a distinctively elevated viscosity of 21.76 ± 1.66 Pa s, while values of 18.52 ± 1.08 Pa s and 16.78 ± 0.84 Pa s, were recorded for LPC and OPFF, respectively. LPI, FPI, FPCD, OPF, and LPF formed an intermediate group with apparent viscosities ranging from 13.92 ± 0.78 Pa s to 15.60 ± 0.69 Pa s. Another group comprising CPC, FPCD and OPC showed viscosities from 12.66 ± 1.55 Pa s to 14.16 ± 0.88 Pa s, without significant differences within this range. The lowest viscosities were measured for CEP, FPC, and CPF, with values between 8.60 ± 0.38 Pa s and 9.99 ± 0.40 Pa s.

Bake loss

3.2.3

Bake loss of the cakes ranged from 10.61 ± 1.56% to 16.31 ± 1.96% among the different formulations (Fig. 3, Table S-2). OF exhibited the highest bake loss (16.31 ± 1.96%). The lowest bake loss was observed for FPF (10.61 ± 1.56%), with LPF also displaying a comparatively low value (12.17 ± 0.41%), and both being statistically different from the majority of samples. Most other samples, demonstrated bake losses between 13.76 ± 0.13% and 14.63 ± 0.61%

Specific volume

3.2.4

The specific volume of the cakes showed distinct differences among the samples, with values ranging from 1.74 ± 0.04 mL/g to 2.21 ± 0.05 mL/g (Fig. 3, Table S-2). The highest specific volume was observed for CEP (2.21 ± 0.05 mL/g), followed by LPF (2.18 ± 0.03 mL/g) and FPF (2.13 ± 0.03 mL/g). FPC and OF also presented relatively high values of 2.00 ± 0.05 mL/g and 2.03 ± 0.02 mL/g, respectively. Intermediate specific volumes were noted for FPCD (1.94 ± 0.03 mL/g), LPC (1.94 ± 0.05 mL/g), CPF (1.93 ± 0.02 mL/g), and CPFD (1.93 ± 0.02 mL/g), with OPC (1.86 ± 0.01 mL/g), OPF (1.86 ± 0.04 mL/g), and OPFF (1.84 ± 0.03 mL/g) showing slightly lower values. The lowest specific volumes were recorded for CPC (1.74 ± 0.04 mL/g), FPI (1.76 ± 0.04 mL/g), and LPI (1.76 ± 0.03 mL/g).

Textural analysis

3.2.5

The results of the textural analysis are displayed in Fig. 3 and Table S-2. The hardness of the cakes ranged from 7.55 ± 0.69 N to 13.01 ± 1.34 N. The highest value was measured for LPF (13.01 ± 1.34 N), followed by CPC (11.75 ± 0.85 N), FPI (10.64 ± 0.74 N), FPF (10.62 ± 0.85 N), LPI (10.61 ± 0.93 N), and OPFF (9.89 ± 1.35 N). CEP, FPC, CPFD, and OPF exhibited intermediate hardness values from 9.12 ± 0.60 N to 9.89 ± 1.35 N. FPCD, OPF, and OF showed lower hardness ranging from 7.55 ± 0.69 N to 8.65 ± 0.67 N, while LPC and OPC had values of 8.11 ± 1.06 N and 8.41 ± 0.43 N, respectively.

Springiness values ranged from 0.81 ± 0.02 to 0.91 ± 0.01. CEP had the highest springiness (0.91 ± 0.01), while the other samples, clustered closely together with values between 0.81 ± 0.02 (FPI) and 0.86 ± 0.03 (CPC).

Cohesiveness values spanned from 0.51 ± 0.03 (LPC) to 0.66 ± 0.02 (CEP). All plant-based cakes clustered within a cohesiveness range of 0.51 ± 0.03 (LPC) to 0.60 ± 0.02 (CPC), while only the CEP cake exhibited a distinctly higher value.

Discussion

4

This study aimed to evaluate the compositional and functional properties of a diverse set of plant protein ingredients in comparison with whole egg powder (CEP), with the objective of providing detailed insights into their potential as egg replacers in bakery applications. The results highlight both the strengths and limitations of the different plant sources and offer practical guidance for their application in pound cake formulations.

Prior to incorporating the diverse ingredients into the cake formulation, compositional analysis is essential for determining their optimal application. Consistent with previous reports, CEP exhibited a balanced profile characterised by high protein and fat contents, accompanied by low carbohydrate levels (Halm et al., 2025). The fourteen ingredients investigated can be classified into three categories: protein isolates (FPI, LPI), protein concentrates (FPC, FPCD, LPC, CPC, OPC), and protein flours (FPF, LPF, CPF, OPF, OPFF, OF). The protein isolates demonstrated the highest protein concentrations (>78 g/100g), with minimal fat and undetectable carbohydrate levels. Multiple studies have examined the nutritional composition of LPI and FPI. Lentil protein isolate typically contains high protein content, ranging from 82% to 90% depending on the extraction method, alongside low levels of carbohydrate (5-7%) and fat (6-8%) (Shevkani et al., 2024; Ghasemi et al., 2025; Miranda et al., 2022). Similarly, commercially produced FPI often exceeds 90% protein when purified via isoelectric precipitation is applied (Badjona et al., 2023; Stone et al., 2024; Thomsen et al., 2025). Among the protein concentrates, intermediate protein contents (53–67 g/100g), varied fat profiles, and moderate carbohydrate levels were observed, reflecting their partial enrichment during processing. Previous studies indicate that deflavouring has minimal impact on the nutritional composition of FPC, with a similar trend noted for CPFD (Halm et al., 2025). In contrast, protein flours generally displayed the lowest protein contents (14–31 g/100g) and the highest carbohydrate levels (up to 72 g/100g), accompanied by relatively low fat. This composition is representative of the minimal processing steps involved in producing protein flours, which typically consist of milling and sieving whole or dehulled seeds into a fine powder while maintaining much of the original seed's nutritional profile (Mondor et al., 2022). It is important to note that protein contents were determined using a single nitrogen-to-protein conversion factor of 6.25 for all ingredients. This uniform factor was chosen to ensure internal consistency across the diverse set of plant-based ingredients and the egg control and to facilitate comparison with previous studies, where 6.25 is commonly applied (Halm et al., 2025; Ghasemi et al., 2025; Miranda et al., 2022; Badjona et al., 2023; Stone et al., 2024). As a consequence, the reported values should be regarded as method-dependent estimates rather than absolute source-specific protein contents, and they may overestimate the true protein content of the plant-based ingredients relative to source-specific factors (Pferdmenges et al., 2025). Conversely, the calculated carbohydrate fraction (by difference) is likely underestimated, consistent with the interpretation that non-protein components substantially contribute to the functional behaviour of flour-based systems. Overall, these compositional trends underscore the importance of selecting plant protein types according to their macronutrient balance and their anticipated functional role in the pound cake matrix.

Principal component analysis (PCA) (Fig. 4) illustrates relationships between the nutritional properties of the ingredients, their techno-functional behaviour, and their influence on cake characteristics. In this study, PCA was used as an exploratory tool to visualise clustering and co-variation among variables rather than as a predictive model for formulation optimisation. Additionally, these relationships are summarised in the correlation matrix in Figure S-1, which visualises pairwise Pearson correlation coefficients. Protein content exhibited a strong positive association with several functional parameters, including TTA (R-value: 0.79; p < 0.05), emulsifying properties (R-value: 0.70; p < 0.05), water-holding capacity (WHC) (R-value: 0.65; p < 0.05), and foam capacity (R-value: 0.68; p < 0.05). Thus, the observed variability in foaming properties among plant-based ingredients can be attributed to differences in protein structure, purity and associated components reported in the literature (Amagliani et al., 2021). Isolates such as LPI and FPI generally yield higher foam capacities due to their elevated protein concentrations, potentially higher surface activity and reduced presence of non-protein constituents, which facilitate rapid adsorption and unfolding at the air-water interface (Amagliani et al., 2021). In contrast, protein concentrates and flours, which typically contain higher levels of carbohydrates, fibre and lipids (Mondor et al., 2022), often display enhanced foam stability but reduced foam capacity, as non-protein components can stabilise or impede foam formation depending on their interactions with the protein matrix. In the present pound cake system, however, high foam capacity measured in simple dispersions did not automatically result in higher specific volume for all ingredients, indicating that expression of foaming functionality is constrained by the high sucrose and fat levels and by interactions with wheat components in the batter (Halm et al., 2024). Additionally, the source of the plant protein further modulates these properties. Previous studies have shown that legume proteins often exhibit flexible molecular structures and favourable surface hydrophobicity that support foam formation, whereas cereal proteins such as those from oat tend to form less stable foams due to more rigid structures and higher levels of interfering polysaccharides or fats (Kumar et al., 2021; Brückner-Gühmann et al., 2018; Huamaní-Perales et al., 2024). Processing steps, such as deflavouring, also modify protein solubility and interfacial behaviour, thereby affecting functional outcomes in foaming systems (Halm et al., 2025).Fig. 4. Principal component analysis (PCA) plot of determined ingredient and respective cake attributes. Representing CEP = whole egg powder; FPI = faba bean protein isolate; FPC = faba bean protein concentrate; FPCD = faba bean protein concentrate deflavoured; FPF = faba bean protein flour; LPI = lentil protein isolate; LPC = lentil protein concentrate; LPF = lentil protein flour; CPC = chickpea protein concentrate; CPF = chickpea protein flour; CPFD = chickpea protein flour deflavoured; OPC = oat protein concentrate; OPF = oat protein flour; OPFF = oat protein flour fine; OF = whole grain oat protein flour.Fig. 4

The protein solubility of the ingredients is directly linked to foam stability (R-value: 0.73; p < 0.05). This relationship was reflected in the cake data, where ingredients with higher solubility and foam stability, such as FPC, FPF and LPF, tended to support higher specific volumes, whereas highly foaming but poorly soluble isolates did not achieve comparable expansion. The high solubility of FPC, FPF, and LPF relative to the purified protein ingredients is primarily attributed to reduced processing-induced denaturation, retention of native protein structures, and the synergistic effects of additional flour constituents on protein-water interactions (Vogelsang-O’Dwyer et al., 2020; Badjona et al., 2023; Stone et al., 2024; Oo et al., 2018). Protein isolates such as FPI and LPI, as well as protein concentrates like CPC and FPCD, demonstrated notably low solubility, most likely as a result of intensive extraction techniques that promote aggregation and denaturation. Conversely, CPF and CPFD showed lower solubility compared to their faba bean and lentil counterparts, an outcome that can be attributed to increased protein hydrophobicity, greater formation of insoluble aggregates, physical interactions with native carbohydrates, and the presence of antinutritional factors known to inhibit protein-water interactions and dispersibility at neutral pH (Vinod et al., 2023; Hong et al., 2024; Patil et al., 2024a). Regardless of processing methods employed, oat protein ingredients consistently displayed the lowest protein solubility. Oat protein is predominantly comprised of globulins (70-80%), with smaller contributions from albumins and prolamins (Senarathna et al., 2024). Globulins are recognized for their limited water solubility, especially at neutral pH, as their molecular structure favours aggregation over hydration. Unlike legume proteins, globulins in oats possess fewer charged and hydrophilic surface residues, thereby hindering water dispersion (Jiang et al., 2015; McLauchlan et al., 2024).

Emulsifying properties are critical for the formation and stability of oil-water systems in cake batter and can influence specific volume, texture and mouthfeel, provided that emulsifying activity is maintained under the high sugar and fat conditions of the batter. In this study, the separation rate was employed as an indicator of emulsion stability, with lower separation rates signifying improved emulsifying stability (Halm et al., 2025). Consistent with expectations, CEP demonstrated superior emulsifying performance that could not be fully replicated by the plant-based samples. Notably, oat flours (OPF, OPFF, OF) exhibited the lowest separation rates among the plant proteins and most closely resembled the emulsifying characteristics of CEP. This emulsifying ability is primarily attributed to oat proteins, especially avenalin, known for its solubility and flexibility, which facilitates stabilisation at the oil-water interface. The physical properties of oat ingredients can be related to their typically high dietary fibre and β-glucan contents reported in the literature, which strongly influence water-binding, viscosity, and gel formation (Kumar et al., 2021; Grundy et al., 2018; Rawal et al., 2023; Shah et al., 2016). Although fibre fractions were not quantified in the present study, the observed behaviour of the oat ingredients is consistent with these literature reports. Furthermore, the literature indicates that β-glucans and other oat polysaccharides can increase the viscosity of the continuous phase, thereby reducing droplet coalescence and enhancing emulsion stability (Kumar et al., 2021; Grundy et al., 2018; Rawal et al., 2023; Shah et al., 2016). Despite their excellent emulsion stability in the model system, cakes made with oat flours did not fully match the specific volume of CEP, suggesting that superior emulsification alone is insufficient in this high-sugar, high-fat matrix and must be complemented by suitable solubility and structure-setting behaviour. Chickpea products were the second most effective plant-based emulsifiers, with CPF displaying promising performance in the model emulsion system and in cakes. This may be due to minimal processing preserving native globulins capable of enhancing emulsification through charge effects and exposure of hydrophobic groups, thus stabilising the oil-water interface (Karaca et al., 2011). In contrast, faba bean and lentil ingredients exhibited poorer emulsifying properties, reflecting the trend that extensively processed products (such as FPI and LPI) lack the ability to create strong, elastic or viscoelastic films at the interface, as observed with chickpea and oat proteins (Lam and Nickerson, 2013).

A potential explanation for the enhanced emulsifying characteristics of oat proteins may relate to their sulfhydryl (SH) group content, which showed a strong negative correlation with separation rate (R-value: −0.76; p < 0.05). These relationships indicate that SH content is relevant for interfacial stabilisation in this system, but SH measurements in dilute dispersions provide indirect evidence and do not capture all reactions occurring during baking. Nevertheless, the contribution of SH-mediated disulfide bonding to cake structure will also depend on the thermal and compositional environment of the batter, so SH content should be interpreted as a potential rather than a guaranteed advantage. Thus, the negative correlation between SH content and separation rate is interpreted as an association consistent with the proposed mechanism, rather than as proof that SH content alone determines emulsifying performance in the cake system. Elevated SH group concentrations, as observed in CEP and oat-derived ingredients, are associated with increased molecular flexibility and the capacity for proteins to form disulfide bonds, which are essential for developing elastic and stabilising interfacial films at oil-water boundaries (Wu et al., 2021; Klose and Arendt, 2012). In contrast, ingredients with lower SH group levels, such as isolates and certain concentrates, typically undergo extensive processing that induces protein aggregation and reduces the availability of accessible SH groups, thereby limiting protein rearrangement and network formation (Wu et al., 2021). De Angelis et al. (De Angelis et al., 2024a) reported total SH group values of 46.65 μmol/g for chickpea, 44.86 μmol/g for faba bean, and 53.40 μmol/g for lentil protein ingredients, highlighting that SH group concentration may vary due to factors such as amino acid composition and genotype in comparable pulse protein ingredients. Moreover, extraction procedures significantly impact SH content, as processing can induce protein unfolding and aggregation, particularly during heat-induced gelling (Yang et al., 2021). Milder processed protein flours (FPF, LPF, CPF, CPFD) exhibited higher SH group concentrations compared to their respective concentrates and isolates, supporting the notion that milder processing preserves native protein structures and reactive groups.

Water-holding capacity (WHC) and oil-holding capacity (OHC) are critical techno-functional attributes in bakery applications (Halm et al., 2025). The results reveal pronounced differences between ingredient types, with FPI and LPI exhibiting exceptionally high WHC, whereas concentrates and flours presented lower values. High WHC is typically associated with a greater abundance of polar and hydrophilic amino acid side chains and an increased protein surface area, as reported in the literature for these protein sources (Badjona et al., 2023; Alonso-Miravalles et al., 2019). In this pound cake formulation, very high WHC in isolates coincided with high batter viscosity and, for some ingredients, reduced specific volume, indicating that an optimal rather than maximal WHC is desirable. Several studies have shown that protein isolates generally possess higher WHC compared to the corresponding flour forms, as non-protein constituents such as starch granules, fibres, and lipids may impede water penetration and binding (Aryee and Boye, 2017; Ma et al., 2022). OHC values were generally less variable, but several ingredients (FPCD, LPC, CPF, OPF, OPFF, and OF) matched or approached the OHC of CEP. The well-balanced hydrophilic and hydrophobic regions of CEP can be mimicked by plant proteins given appropriate concentration or modification, thereby reproducing the microstructural and functional behaviours of eggs with respect to oil retention and emulsification (Buhl et al., 2019). FPCD achieves high OHC, which has been linked in previous work to its specific conformation and amino acid profile (Halm et al., 2025). LPC provides beneficial functionality in bakery systems owing to surface-active proteins and small functional particles (Jarpa-Parra et al., 2017), and CPF has been reported to exhibit superior fat retention, which has been associated with higher proportions of nonpolar amino acids in its protein fraction (Patil et al., 2024b; Sudarsanan et al., 2023). Additionally, oat protein flours contain both soluble and insoluble proteins which, when combined with starch-lipid complexes, further enhance oil-holding capacity (Shah et al., 2016; Laursen et al., 2024).

The minimum gelling concentration is a widely used parameter for evaluating the ability of proteins to form gel networks. Among the samples analysed, oat flours (OPF, OPFF, OF) exhibited the most favourable gelling capacity, which reflects the combined action of starch and protein under aqueous conditions rather than protein alone. However, this low minimum gelling concentration did not translate into optimal crumb structure in the sucrose- and fat-rich batter, as indicated by their only intermediate specific volume. In such systems, multiple flour components contribute to network formation, so a lower nitrogen-to-protein conversion factor would decrease the calculated protein content and increase the apparent carbohydrate fraction, further emphasising the role of non-protein material in gel formation (Sun et al., 2023). In contrast, several protein flours and concentrates (LPF, CPC, CPF, OPC) required higher concentrations to form gels, likely due to dilution of gelling-active proteins or interference from residual starch and dietary fibre (De Angelis et al., 2024a). Previous work on FPC and FPCD also showed that minimally processed fractions better replicate the minimum gelling concentration and network properties of CEP, underscoring the importance of preserving native protein structure and its interactions with co-existing starch (Halm et al., 2025).

When integrating alternative protein ingredients into a cake system, it is essential to consider their techno-functional properties. The selection and optimisation of specific plant-based proteins for cake production must be guided by their capacity to deliver these functionalities within the complex matrix of bakery products, thereby enabling the development of cakes with desirable texture, volume and quality while supporting nutritional and sustainability goals. Key parameters such as initial structure setting temperature (T_i_), storage modulus at structure setting (G'i) and apparent viscosity reflect the matrix's ability to retain air and develop structure throughout mixing and baking. It should be noted that, although mixing time, speed and equipment were standardised across treatments, small differences in batter aeration cannot be completely excluded and may have contributed to variability in cake parameters, such as batter viscosity and specific volume, in addition to ingredient-related effects. Inclusion of plant protein ingredients led to pronounced variations in apparent batter viscosity compared to CEP. Batters containing LPF, CPFD, FPF, and OF demonstrated notably higher viscosity, likely attributable to the combined effects of starch, dietary fibre, and protein network formation typically associated with these less processed protein flour ingredients as reported in the literature (Aghababaei et al., 2024; Badia-Olmos et al., 2023; Han et al., 2025; Singh et al., 2015), rather than the protein alone. Additionally, for some formulations, this high viscosity likely limited bubble expansion, contributing to lower specific volume despite favourable water binding. Interestingly, CPF exhibited similar apparent viscosity to CEP, despite possessing a nutritional composition analogous to CPFD. This may be explained by CPF's lower water-holding capacity, which arises from protein modification, burying of hydrophilic groups, and removal of interfering components through deflavouring (Halm et al., 2025). Changes in batter viscosity are critical, as they influence air incorporation, bubble stability, and ultimately affect cake expansion and crumb texture during baking (Halm et al., 2024).

The determination of rheological parameters via temperature-dependent measurements provides valuable insights into the structure-setting dynamics of cake batters. The initial gelling temperature (T_i_) reflects the point at which significant protein denaturation, starch gelatinisation, and the onset of elastic network formation occur during heating. Batters formulated with CEP displayed a structure-setting point similar to those containing FPF and LPF, suggesting comparable thermal stability and setting profiles. Lower T_i_ values observed with protein isolates and concentrates may result from their altered, more aggregation-prone protein structures, minimal starch content, and the absence of naturally protective components found in flours, leading to rapid heat-induced protein network formation (Asen and Aluko, 2022; Aslam et al., 2026; Gul et al., 2025). Additionally, the reduced T_i_ in cake batters containing oat ingredients is attributed to the early gelatinisation of oat starch (∼64 °C), softer and more thermolabile gelling behaviour of oat proteins, and the presence of a matrix rich in lipids and beta-glucans as reported for oat flours in the literature (Sergiacomo et al., 2025; Brückner-Gühmann et al., 2019; Eazhumalai et al., 2025). The storage modulus at structure setting (G'i) quantifies the batter's elastic character at this gelling threshold. Higher G'i values, such as those observed for FPF and LPF, indicate the development of a stronger, firmer protein-starch network upon heating, likely reflecting the combined contribution of functional proteins, residual starch, and dietary fibre that are typically present in these less processed flour ingredients, rather than protein alone. Typically, lower G'i values, as seen in the CEP system, reflect a more flexible and less rigid network, which promotes greater expansion and a lighter cake texture in egg-based cakes (Inanlar and Altay, 2024). Conversely, lower G'i in the remaining plant ingredients was associated with reduced specific volume in the final product (R-value: 0.52; p < 0.05), suggesting that weaker networks were less able to support expansion.

Specific volume is a key indicator of cake quality. Cakes produced using CEP exhibit the highest specific volume, consistent with the well-established ability of egg proteins to stabilise air cells and promote batter expansion during baking (Halm et al., 2024, 2025). Correlation analysis provided insights into essential parameters influencing specific volume: initial structure setting temperature (T_i_, R-value: 0.67; p < 0.05) and protein solubility (R-value: 0.82; p < 0.05). These relationships support the relevance of the solubility and micro-baking measurements for predicting cake performance in this system, as higher solubility and delayed structure-setting were associated with improved specific volume. Nonetheless, these correlations remain associative and are interpreted in the light of simplified assays used, and do not establish direct causal relationships. Moreover, the fact that some ingredients with high foaming or emulsifying capacity (e.g. isolates) did not yield high specific volumes underscores that such functional advantages are only partially expressed under the conditions of this high-sugar, high-fat pound cake formulation. However, these favourable properties are apparent in cakes containing FPF and LPF, which achieve the highest specific volumes following CEP. A higher structure setting temperature allows prolonged expansion and enhanced air incorporation, reducing the risk of premature collapse. Meanwhile, greater protein solubility contributes to stabilisation of air bubbles during baking, as soluble proteins migrate more effectively to the air-water interface. Accordingly, when egg replacers possess both of these techno-functional characteristics, their performance in cake batters more closely mirrors that of eggs (Halm et al., 2024; Yazici and Ozer, 2021). However, for several plant-based formulations, gains in specific volume were accompanied by less favourable textural attributes, such as increased hardness or reduced cohesiveness compared to CEP, underscoring that specific volume alone does not capture overall cake quality. Interestingly, OF yields cakes with a favourable specific volume without displaying these beneficial techno-functional properties. The high fibre and β-glucan commonly reported for whole oat flour may account for superior water absorption and retention, supporting improved gas retention and expansion during baking. Additionally, oat flour contains lipids, non-gluten proteins, and pentosanes, which can stabilise the batter matrix and form viscous, gel-like solutions that favour increased cake rise (Popa et al., 2021). Despite its relatively high specific volume, OF still differed from CEP in texture-relevant attributes such as springiness and cohesiveness, indicating that improved gas retention does not fully compensate for the absence of egg multifunctionality.

Texture analysis revealed differences in hardness, springiness, and cohesiveness across the cake samples, highlighting the impact of ingredient composition and processing on crumb structure. Cakes formulated with CEP exhibited relatively low hardness, alongside high springiness and cohesiveness, characteristics typical of the soft, elastic, and integrated crumb associated with traditional egg-based cakes (Halm et al., 2024, 2025). Comparable hardness values were observed in cakes containing FPC, FPCD, OPF, OPFF, and OF. Halm et al. (2025) attributed the comparable hardness of FPC and FPCD to the presence of soluble fibres and the formation of amylose-lipid complexes, which may modify starch gelatinisation. Similarly, the hardness observed in oat-based ingredients could result from the gel-forming capability and structural contributions of oat proteins and associated carbohydrates in the flour. OPF, OPFF, and OF had the lowest minimum gelling concentration. Oat proteins, particularly globulins and avenins, can establish networks or gels under cake-baking conditions, which entrap air and retain water, thus contributing to comparable textural parameters (Sergiacomo et al., 2025; Cao et al., 2022; De Angelis et al., 2024b). Nonetheless, these formulations often exhibited lower springiness or cohesiveness than CEP, reflecting a firmer and less elastic crumb despite acceptable hardness. Springiness was highest in the CEP control, with most plant-based substitutes unable to replicate the effects of egg, except for CPC. This outcome may relate to similar foam stability, water- and oil-holding capacity compared to CEP, enhancing the formation of a well-balanced protein-starch-lipid matrix able to retain small gas cells and maintain an elastic, springy crumb (Patil et al., 2024b; Caldeira et al., 2025; Sohaimy et al., 2021). Even for CPC, however, the improvement in springiness did not extend to all texture dimensions, and some increase in hardness relative to CEP remained. Elevated cohesiveness more similar to CEP was only evident in cakes containing chickpea products (CPC, CPF). Cohesiveness, reflecting the internal bonding of the cake crumb, was reduced in other plant-based ingredient systems, particularly those with higher protein content and less native structure, where protein denaturation and aggregation can disrupt matrix integrity. Yet these chickpea-based cakes still differed from the egg control in other attributes such as specific volume, illustrating that matching one texture parameter does not equate to full equivalence with eggs. Taken together, the cake results indicate that several key aspects of egg functionality remain difficult to reproduce with single plant ingredients, concurrent high aeration and emulsion stabilisation, maintenance of soft yet elastic and cohesive crumb, and robust structure-setting over a broad processing window. In our system, plant proteins could approximate selected aspects of this multifunctionality, but not the full combination achieved by whole egg powder. It should be emphasised that these assessments are based solely on instrumental measurements of specific volume and texture. No trained panel or consumer sensory evaluation was conducted, so practical acceptability can only be inferred and should be confirmed in future studies.

The techno-functional assays were performed in simplified model systems that do not fully reproduce the complexity of high-sugar, high-fat pound cake batters. Foaming and emulsification data should therefore be viewed as indicators of intrinsic interfacial behaviour rather than direct predictors of cake performance, and the observed associations with specific volume and texture are interpreted with caution. Additionally, it should be noted that amino acid composition and protein surface hydrophobicity were not experimentally determined in this study. The discussion of these properties is based on published literature and mechanistic considerations rather than direct measurements in our samples. Future work incorporating in-batter or in-process measurements would further clarify refine these links.

Conclusion

5

The findings of this study demonstrate that plant protein ingredients, faba bean, lentil, chickpea and oat, provide distinctive techno-functional properties that partially fulfil the roles of egg in pound cake formulations. Although none of the plant-based alternatives completely replicate the structural and functional attributes of eggs, strategic selection and processing can yield formulations that approach the egg control in specific attributes such as specific volume or selected texture parameters, while still exhibiting compromises in others. Differences in foaming, emulsifying, water- and oil-holding and gelling capacities among isolates, concentrates and flours highlight the importance of understanding compositional and molecular characteristics to achieve desirable instrumental baking performance. Importantly, the comparative design used here, including ingredient-specific optimal inclusion levels and the absence of strict protein or functional-solids normalisation across all formulations, means that the effects observed reflect combined changes in protein and non-protein components and should not be interpreted as a pure protein-functionality comparison.

Notably, faba bean and lentil flours delivered high protein solubility, improved specific volume, and acceptable textural outcomes, approaching egg-based controls for certain parameters, while chickpea flour stood out for its contributions to cake cohesiveness and springiness. Oat flours, whose fibre and β-glucan contents are known from the literature to promote emulsification and gel formation, likely contributed to the cake matrices with favourable moisture retention and structure. Nevertheless, most plant protein ingredients resulted in cakes with greater hardness and/or lower springiness or cohesiveness than the conventional egg-containing formulation, underscoring that improvements in one quality dimension often come at the cost of others and that full functional equivalence to eggs, particularly the simultaneous delivery of high volume, softness, elasticity and cohesiveness, was not achieved.

To address these limitations, future research should explore synergistic combinations of the most promising plant-based egg replacers, as well as targeted modification strategies (e.g. enzymatic or process treatments), aiming to better approximate the multifunctionality of CEP while explicitly balancing trade-offs between volume, texture and nutritional profile. Such strategies may help to approach the technological performance of eggs but also match or exceed the biological value of CEP's amino acid profile, supporting both potential sensory quality and nutritional adequacy in plant-based bakery products.

Overall, this research advances practical understanding of plant protein ingredient selection for egg replacement in baked goods, emphasising the significance of compositional characterisation and functional evaluation in guiding formulation innovation. Unlocking the full potential of plant proteins therefore lies not in single ‘one-to-one’ substitutes, but in rationally designed systems that manage the inherent trade-offs between functionality, quality and sustainability.

Author contributions

Conceptualisation, E.K.A. and E.Z.; Methodology, J.H.; Formal Analysis, J.H., Resources E.K.A.; Data Curation, J.H.; Writing-Original Draft Preparation, J.H.; Writing-Review and Editing, E.K.A., E.Z. and L.N.; Visualization, J.H.; Supervision, E.K.A., and E.Z.; Project Administration, E.K.A. and E.Z.; Funding Acquisition, E.Z. and E.K.A. All authors have read and agreed to the published version of the manuscript.

Data availability statement

All data is contained within the article.

Funding

The research for this publication has been undertaken as part of the GIANT LEAPS project. This project has received funding from the European Union’s HORIZON EUROPE research and innovation programme under grant agreement N° 101059632.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.