The use of protein binders and sorghum crisps as potential ingredients in a cereal bar for dogs

This study aimed to evaluate the inclusion of different protein binders and sorghum crisps in cereal bars for dogs and their effect on sensory properties, product texture, and dog preference. Fifteen cereal bars were developed in which three crisp sources (rice crisp, white and red sorghum crisp) and five sources of binders (corn syrup, spray dried plasma, gelatin, albumin, and egg product) were evaluated. An interaction effect between binder and crisp sources was found for textural properties ( p <.05). A total of 103 volatile compounds were identified and semi-quantified in the cereal bar samples, with aldehydes being the most represented. Unlike crisp source, protein binders played a major role on sensory properties and impacted the dog's preference. This study suggests that sorghum crisps and protein binders may be used in cereal bars for dogs; however, considerations regarding sensory attributes and dog's preference should be taken to maximize product acceptance. This is the first study to report information regarding the use of novel ingredients in a cereal bar application for dogs. The findings observed wherein provide a compre-hensive understanding about product development and the impact of ingredients on final product quality, sensory properties, and animal preference.


| INTRODUCTION
The pet food industry was worth over $69 billion in 2017 with a market growth projected to be over $72 billion for the following year (APPA, 2018). This rapidly growing market is driven by the development of new products, which involve the addition of new ingredients and food forms. The pet food segment can be nutritionally segregated into two major categories: complete & balanced, and snacks & treats. Although complete & balanced products compose the greatest share of the market sales, snacks & treats represent an important part of the industry. More than ever, pets are considered part of the family, and these products are offered as a demonstration of affection and love.
The humanization of pets has greatly impacted the pet food industry trends. Human food trends are being translated to the pet food industry as pet owners demand pet foods that reflect their own dietary choice. Trends toward more healthy and natural food have gained popularity in the pet food industry (Sprinkle, 2018). Cereal bars were introduced in the human food industry several decades ago as a healthy snack alternative (Bower & Whitten, 2000). These products remain popular, and they may represent a potential alternative to healthy treats for dogs. However, a previous study revealed that the chemical composition of some cereal bars in the market was similar to confectionary (Boustani & Mitchell, 1990). This may not be perceived as healthy by pet owners, and it could exclude the product from this claim.
Most cereal bars are composed of several ingredients that are bound by the addition of sugar syrups. This amplifies the levels of soluble carbohydrates in the final product. Although carbohydrates are well utilized by dogs, many pet owners perceive sugars as unhealthy and seek protein-rich food forms for their pets. Proteinaceous ingredients such as egg white and bovine plasma have been used in the gluten-free industry as potential binders (Crockett, Ie, & Vodovotz, 2011;Furlan, Padilla, & Campderr os, 2015;Han et al., 2019), and may be a replacement for sugar syrup for manufacturing cereal bars. Rice crisps are also commonly used in cereal bar recipes. The replacement of rice crisps for nutraceutical ingredients such as sorghum may add value to the product. Sorghum is rich in phytochemicals that are known to have antioxidant and antiradical activities (Hagerman et al., 1998). The replacement of sugar syrup and rice crisps for soluble animal protein binders and sorghum crisps, respectively, may improve the nutritional quality of cereal bars and add value to the product for the pet market. However, to validate the use of these new ingredients in cereal bars, it is crucial to evaluate how these replacements (rice crisp vs. sorghum crisp and sugar syrup vs. protein binders) are accepted not only by the dog, but also by the owners as they make the purchase decision. To date, there are no studies evaluating acceptability/preference of cereal bars by dogs, and the use of protein binders and sorghums crisps as potential ingredients for this application. Thus, the objectives of this study were to (1) evaluate the effect of different proteinaceous binders and sorghum crisps on the texture and volatile compounds profile of cereal bars for dogs, (2) describe the sensory profile of the cereal bars using a highly trained human panel, and (3) to evaluate the effect of these ingredients on dog preference.

This research was approved by Kansas State University Institutional
Review Board protocol #5930 and Institutional Animal Care and Use Committee under protocol #3722.

| Cereal bar production
This experiment was conducted as a 3 Â 5 factorial arrangement for simultaneous evaluation of three sources of crisp (rice crisp, RC; white sorghum crisp, WSC; and red sorghum crisp, RSC) and five sources of binder (corn syrup, CS; spray dried plasma, SDP; gelatin, GL; albumin, AL; and egg product, EP). The RC and the CS were used as positive control from the crisp and binder effect, respectively. The WSC and RSC used in this study were produced using a single screw extruder (model E525, ExtruTech, Inc., Sabetha, KS) from a previous experiment at Kansas State University (Pezzali, Suprabha-Rah, Siliveru, & Aldrich, 2020). The RC was sourced from local manufacture (Cereal Ingredients Inc., Leavenworth, KS). The SDP (Innomax Porcine Plasma), GL (Pro-Bind Plus 50), and AL (Innomax MPI) were acquired from Sonac ® (Maquoketa, IA), and the EP (Ovabind-RSD 80) was acquired from IsoNova ® (Springfield, MO). Last, CS (Light Corn Syrup, Kroger ® ) was acquired from a local grocery market.
Before mixing, the WSC and RSC were manually ground to decrease particle size. The agglomerating syrup was added to the dry ingredients, and the mixture was stirred until the crisps were uniformly coated. The syrup was composed of corn syrup or one of the protein sources. Protein sources required hydration prior addition into the mix and were prepared as follows: SDP (1:1.76 ingredient/water, wt/wt), GL (1:2.5 ingredient/water, wt/wt), AL (1:1.18 ingredient/ water, wt/wt), and EG (1:1.47 ingredient/water, wt/wt). Protein sources required different processing conditions to create a consistent agglomeration syrup. The GL and AL were hydrated under heat. An attempt to hydrate SDP and EP under heat was performed, but these protein sources started to cook; therefore, this step was not implemented. Corn starch was added to all proteinaceous agglutination syrup with exception of the SDP. The final mixture was transferred to a cookie sheet covered with parchment paper and baked for 20 min at 163 C in a convection oven (MEA 21-93-E; Garland Commercial Industries, PA). After baking, the dough was cut into approximately 4 Â 4 cm square pieces. Treatments that contained protein ingredients in the agglomerating syrup were dried overnight at 55 C in a convection oven (212041, HotPack, PA) to achieve a moisture content below 10%. Experimental treatments were produced as described above over three replicate-days.

| Nutritional analysis
For each dietary treatment, a sample of 50 g from each day of production was composited and ground through a 1-mm screen in a fixed blade laboratory mill (Retsch, type ZM200, Haan, Germany

| Textural properties
Hardness and toughness of cereal bars were assessed using a Shimadzu EZ-SX Texture Analyzer (Shimadzu Corporation, Kyoto, Japan). Five cereal bar square pieces were randomly selected for each replicate, totaling 15 samples per treatment. A compression test was performed using a toothed pushrod B probe at a speed of 1.67 mm/s. Hardness was defined as the highest peak fracture force. Toughness was defined as the total energy required to break the sample as it was calculated as the total area under the fracture curve.

| Preference ranking test
The preference test took place at the Large Animal Research Center (LARC) at Kansas State University where 12 castrate Beagle dogs were used. Dog's preference was evaluated under a preference ranking procedure according to Li et al. (2018). Four preference raking tests were performed, and each one consisted of 10 days (5 days of acclimation and 5 days of data collection). Each dog was presented simultaneously with five cereal bar treats in a rubber puzzle toy (Kong ® ). Dogs were allowed to smell the Kongs then each was placed randomly in the left corner of the 1.5 m Â 1.5 m test room. The order in which the cereal bar was extracted and consumed by the dog was considered as the preference ranking order. This ranged from 1 to 5, where 1 was considered the most preferred, and 5 the least preferred.
First, the effect of protein source on dog's preference was evaluated. Three ranking tests were performed in which the treatments having the same crisp source and differing in the binder source were compared. Second, the effect of crisp source on dog's preference was assessed using a modified preference ranking test (Li et al., 2017). The SDP protein source was selected for evaluation. Three cereal bar treats produced with the SDP but differing in crisp source were presented to the dogs. Preference was assessed as described above.

| Descriptive analysis
The 15 cereal bars were evaluated by five highly trained panelists from  The cereal bar samples were stored at À18 C and thawed overnight at 4 C a day before the sensory evaluation. For aroma evaluation, 5 g of each ground sample was served in medium snifters, covered with watch glasses. Three pieces of 2 cm Â 2 cm square cereal bars were served in Ziploc bags (S.C. Johnson & Son, Inc. Racine, WI) for appearance, flavor, and texture evaluation. All of the samples were labeled with random, three-digit blinding codes.

| Volatile compounds profile
The isolation, tentative identification, and semi-quantification of the volatile compounds were performed on a gas chromatograph (Shimadzu GC-2010 Plus, Kyoto, Japan), coupled with a Shimadzu mass spectrometer detector (GCMS-QP2020, Kyoto, Japan). First, volatile compounds were extracted from samples using headspace solid phase micro extraction (HS-SPME) according to a modified version of Koppel, Adhikari, and Di Donfrancesco (2013). The samples were ground in a coffee grinder for 10 s, and a 0.5 g of ground sample was weighed into a 10 ml screw-cap vial equipped with a polytetrafluoroethylene/silicone septum (Supelco, Bellafonte, PA). Exactly 0.99 ml distilled water was added to the ground sample in the vial, and 10 μl of 100 ppm 1,3-dichlorobenzene (Sigma Aldrich, St. Louis, MO) was added as the internal standard. Each sample was incubated at 50 C for 1 min before extraction. A 50/30 μm divinylbenzene/ carboxen/polydimethylsiloxane fiber (Supelco, Bellafonte, PA) was exposed to the sample headspace for 20 min at 50 C for volatile compounds extraction. The volatiles were desorbed from the SPME fiber coating in the injection port at 240 C for 3 min in splitless mode. The GC-MS system was equipped with an SH-Rxi-5Sil MS Crossbond ® column (Shimadzu, Tokyo, Japan; 30 m Â 0.25 mm Â 0.25 μm film thickness), which was heated from 40 to 240 C. The identification of the compounds was done using NIST library version 14. All analyses were run in two replicates. 3 | RESULTS

| Chemical composition
The chemical composition of cereal bars is reported in Table 1. Cereal bars produced with proteinaceous agglutination syrups had CP content an average 19.72% greater than those produced with corn syrup, while NFE content was reduced by 23.38%. Minor differences were observed for fat, CF, and DM.
The hardness and toughness of the cereal bars was affected by an interaction between the binder and crisp source ( Table 2). The CS-RC presented the highest toughness value followed by CS-RSC (p < .05). Compared to these, the EP-RC, EP-WSC, EP-RSC, AL-RC, AL, RSC, SDP-WSC, and SDP-RSC treatments exhibited lower toughness (p < .05). The highest numerical value was observed for GL-WSC while the lowest hardness was observed for CS-RSC.

| Descriptive sensory analysis
All the cereal samples had low to medium levels of brown color (score range: 4-7) and medium to a high level of roughness of surface (score range: 7-10; Table 3). Samples made with RC + AL and RSC + AL exhibited the darkest brown color among the 15 treatments, while samples made with EP had the lightest brown color. The WSC + AL sample was found to have the roughest surface, on the other hand, samples produced with GL appeared to have a smoother surface than other samples.
The 15 samples exhibited a high level of roughness of mass, particles amount, and had medium particle size (Table 3). Furthermore, all treatments had extremely high firmness (Table 3), except for the two treatments made with CS. The cohesiveness of mass was low in all samples (score range: 0-3).
The aroma intensities were low for the attributes found among the samples (score range: 0-5). Vitamin, grain complex, cardboard, heated oil, toasted, nutty, and sweet aromatics were found in all the samples (Table 4). Cereal bars produced with CS did not exhibit butyric nor stale aromas, which were commonly found in the other samples. Samples made with AL had a relatively stronger butyric aroma than other treatments.
Similar to aroma intensities, the flavor intensities were relatively low for the attributes found in all the samples (score range: 0-6.5).
Only five flavor attributes were scored for the 15 treatments including dry fruit complex, nutty, vitamin, grain, and cardboard (

| Preference ranking test
No signs of refusal were observed throughout the four different preference tests performed (  WSC bars made with AL and SDP were highly correlated to aromas of toasted, cardboard, barnyard, butyric flavor, and salty taste.

| Drivers of preference-sensory attributes
WSC made with EP was located near the center of the plot indicating that the sample was not strongly correlated with the differentiating sensory attributes (Figure 1). Dogs preference for the cereal bars produced with WSC was toward WSC + AL and WSC + SDP samples.
This result was following the preference ranking test, in which WSC + SDP was preferred over WSC + EP by dogs. The preference for WSC cereal bars was driven by brown color, roughness of surface, and heated oil.
The PCA of cereal bars produced with RSC as the crisp source can be found in Figure 2.   (Tables S1-S3). Aldehydes were the major volatile class followed by alcohols and alkanes.

| Relationship between sensory attributes and volatile compounds of cereal bars
Possible associations between sensory attributes identified in the cereal bars and volatile compounds were found by PLSR analysis (Figure 5).
Some sensory attributes were highly correlated with the volatiles.  (Figure 6).

| DISCUSSION
The objective of our study was to evaluate the effect of protein binders, and white and red sorghum crisps in cereal bars on sensory properties, textural properties, and dog preference. There is a lack of research regarding processing conditions and animal preference of snacks and treats. The use of alternative ingredients in cereal bars to enhance their protein content has been investigated. Specifically, peanut flour, soy flour, and mesquite cotyledon (Escobar, Estevez, Tepper, & Aguayo, 1998;Estevez, Escobar, & Ugarte, 2000) and black and red beans (Maurer et al., 2005) were included in cereal bars to increase their protein content. However, this is the first study to investigate the replacement of corn syrup by a proteinaceous ingredient in a novel dog treat application. The results reported herein suggest that it is possible to enhance the CP content of cereal bars by using either AL, GL, SDP, or EP as a replacement for CS without compromising product integrity and animal acceptance. To form an agglomeration syrup that effectively binds the dry ingredients, hydration of protein sources was performed differently according to their protein type. Each protein source has a unique amino acid composition, sequence, and molecular weight, which have a direct impact on  (Table A2) F I G U R E 6 Cereal bars were successfully produced with three different crisp sources (RC, rice crisp; RSC, red sorghum crisp; and WSC, white sorghum crisp) and five sources of binders (corn syrup; spray dried plasma; gelatin; albumin; and egg product) F I G U R E 5 Plot of Partial Least Square Regression (PLSR) analysis of the volatile compounds (Xmatrix) and sensory attributes (Ymatrix), 15 cereal bars were shown in green. Volatile compounds are listed as ID numbers (Table A2) applications due to their foam-stabilizing activity. This was the first attempt to use SDP, GL, AL, and EP as binder sources in cereal bar applications. Our study indicated that these protein sources can effectively work as binding agents for cereal bars. It was not our intention to investigate the best inclusion level of each protein source, but rather to evaluate the potential of each protein as binders. The inclusion level of the protein source used in our study may serve as a starting point for future research aiming to investigate the adequate inclusion level of the protein of interest to maximize product quality and animal acceptance. Processing conditions such as baking time and temperature should also be investigated as each protein source may have different processing requirements to perform at its best.
The use of WSC and RSC were also evaluated as replacements for RC. The sorghum crisps used in this research were produced from a previous study (Pezzali et al., 2020) while the rice crisp was acquired from a commercial source. The rice crisps were denser (348 g/L) compared to the white and red sorghum crisps (107 and 92 g/L, respectively). This compromised aggregation of ingredients due to a lower surface area for contact. To include the same mass of rice and sorghum crisps in the recipe, they had to be manually ground (crushed) to decrease particle size, and improve aggregation. Thus, extrusion conditions for production of sorghum crisps may be adjusted in future (e.g., decrease specific mechanical energy) to decrease expansion and increase the density of the product to improve their use as an ingredients for cereal bars. A recent study showed that extruded sorghum flour can promote health benefits in obese rats (Arbex et al., 2018).
Extruded sorghum crisps might provide similar results in obese dogs.
Thereby, they are promising ingredients to be added in a "healthy" Textural characteristic of the product is an important aspect to be considered as it can influence product quality and animal acceptance and preference. In general, the addition of proteins in bread application increase product hardness (Furlan et al., 2015). For example, the inclusion of egg albumin in a gluten-free bread formula improved first bite and masticatory hardness (Toufeili et al., 1994). Although the baking science behind bread production is different than the mild baking step used for the development of cereal bars, we still expected that the use of protein binders would increase hardness compared to sugar syrup. In the current study, hardness was influenced by the interaction effect of binder and crisp source. The inclusion of RC and GL likely resulted in harder products. Other ingredients besides binder and crisp sources could have influenced product hardness as cereal bars were not formulated with fixed amounts of each ingredient. The interaction effect between the binder and crisp source on textural products of the cereal bars may be due to ingredient interaction and its influence on the product matrix, to different inclusion levels of some ingredients, and also to differences in the size of samples used for the analysis. Thus, we cannot draw accurate conclusions solely based on crisp and protein sources. Most of the hardness values observed in our study were similar to those reported by Torres et al. (2011). These authors obtained hardness values between 9.4 and 19 kg for cereal bars. The toughness of cereal bars was assessed as the total energy required to break the sample. The cereal bars formulated with CS and RC showed a greater toughness value compared to the other treatments. The small particle size of RC combined with the binding effect from CS probably resulted in higher aggregation of ingredients, thereby greater energy was required to disintegrate the sample.
The preference ranking test was used to assess the dog's preference. In the protein source evaluation, dogs showed a preference for SDP compared to EP and GL in cereal bars produced with WSC and RSC, respectively. In this test, the aroma is the first factor driving the animal's choice. However, dogs have the chance to associate the aroma profile of the sample with the texture and flavor after the treats are presented for 5 days. The PCA revealed a low correlation between dog's preference for cereal bars produced with either WSC or RSC and the volatiles within the samples. Nevertheless, the lack of a strong correlation does not imply that the volatile profile of the cereal bars did not play a role as a driver of dog preference. Although the preference ranking test has been previously validated (Li, Smith, Aldrich, & Koppel, 2018), it may lack sensitivity to discriminate samples that are overall well accepted by the animals. The fact that the dogs preferred cereal bars produced with SDP over those produced with EP and GL, when WSC and RSC were used as crisps sources, respectively, and the high correlation between cereal bar samples and

| CONCLUSION
The replacement of rice crisps for white and red sorghum crisps did not lead to differences in animal preferences and sensory properties, but the size and density of sorghum crisps should be reconsidered in order to improve product quality. On the other hand, dogs preferred cereal bars produced with spray dried plasma over gelatin and egg product when red and white sorghum crisps were used as the crisp source, respectively. Combinations of different protein binders and crisp sources impacted product texture, thereby the interaction of ingredients should be examined more closely during product development. With regards to the aroma profile, more than 108 volatile compounds were identified in the cereal bars, and their concentration was mostly affected by the proteinaceous ingredients. Thus, proteinaceous binders and sorghum crisps are potential ingredients to replace sugar syrup and rice crisps, respectively, in cereal bar treat application for dogs.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.

Tables A1 and A2
T A B L E A 1 Attributes used in the descriptive sensory analysis of cereal bars White Pepper in water = 5.0 Butyric Aromatics reminiscent of baby vomit; is sour and cheesy.
Kraft Shredded Parmesan Cheese = 5.5 Nutty A combination of slightly sweet, brown, woody, oily, musty, astringent, and bitter aromatics commonly associated with nuts, seeds, beans, and grains.
Le nez du café n. 29 "roasted hazelnuts" = 7.5 Sweet aromatics Aromatics associated with the impression of sweet substances.

Stale
The aromatics associated with wet cardboard that is characterized by a lack of freshness.
Mama Mary's Pizza Crust = 4.5 Flavor Meaty A measure of how much a sample is recognized as distinctly animal muscle tissue.
Canned Swanson Beef Broth = 6.0 Dry fruit complex Aromatics associated with dried brown fruit. Mix of Chopped Prunes, Sun-Maid Figs, and Sun-Maid Raisins = 8.5 Nutty A light, brown, slightly musty aromatic associated with nuts, wheat germ, and certain whole grains.

Vitamin
The aromatics associated with a just opened bottle of vitamin pills.

Grain
The light dusty/musty aromatics associated with grains such as corn, wheat, bran, rice, and oats.
Cereal Mix (dry) = 8.0 Cardboard A flat flavor note associated with cardboard or paper packaging that may be associated with a stale characteristic.

Eggy
Aromatics/flavors associated with cooked whole chicken eggs, with savory, earthy, salty, buttery, and sulfur overtones. May also include sweet, metallic, and cardboard notes.

Sour
The fundamental taste factor associated with a citric acid solution.
0.015% Citric Acid Solution = 1.5 0.050% Citric Acid Solution = 3.5 Salt A fundamental taste factor of which sodium chloride is typical.

Bitter
The fundamental taste factor associated with a caffeine solution.

Texture Firmness
The force required to bite completely through the sample with the molar teeth. Evaluate on first bite down with the molars.

Cohesiveness of mass
Degree to which mass holds together during mastication after 5-7 chews.
Nabisco Triscuit = 2.5 General Mills Cheerios = 7.0 Roughness of mass The degree of abrasiveness perceived when gently manipulating the mass against the palate after 5-7 chews.
General Mills Wheaties = 8.0 Nabisco Triscuit = 11.0 Particles amount Perception of pieces within the sample which do not break down during mastication. Evaluated by chewing 8-10 times, then manipulating the sample with the tongue 3-5 times.
General Mills Wheaties = 10 Particles size Perception of pieces within the sample which do not easily break down during mastication. Evaluated by chewing 8-10 times, then manipulating the sample with the tongue 3-5 times. General