**Diet, Perceived Intestinal Well-Being and Compositions of Fecal Microbiota and Short Chain Fatty Acids in Oat-Using Subjects with Celiac Disease or Gluten Sensitivity**

**Lotta Nylund 1, Salla Hakkola 1, Leo Lahti 2, Seppo Salminen 3, Marko Kalliomäki 4,5, Baoru Yang <sup>1</sup> and Kaisa M. Linderborg 1,\***


Received: 12 August 2020; Accepted: 20 August 2020; Published: 25 August 2020

**Abstract:** A gluten-free diet may result in high fat and low fiber intake and thus lead to unbalanced microbiota. This study characterized fecal microbiota profiles by 16S MiSeq sequencing among oat-using healthy adult subjects (*n* = 14) or adult subjects with celiac disease (CeD) (*n* = 19) or non-celiac gluten sensitivity (NCGS) (*n* = 10). Selected microbial metabolites, self-reported 4d food diaries and perceived gut symptoms were compared. Subjects with NCGS experienced the highest amount of gut symptoms and received more energy from fat and less from carbohydrates than healthy and CeD subjects. Oat consumption resulted in reaching the lower limit of the recommended fiber intake. Frequent consumption of gluten-free pure oats did not result in microbiota dysbiosis in subjects with CeD or NCGS. Thus, the high number of gut symptoms in NCGS subjects was not linked to the microbiota. The proportion of fecal acetate was higher in healthy when compared to NCGS subjects, which may be linked to a higher abundance of *Bifidobacterium* in the control group compared to NCGS and CeD subjects. Propionate, butyrate and ammonia production and β-glucuronidase activity were comparable among the study groups. The results suggest that pure oats have great potential as the basis of a gluten-free diet and warrant further studies in minor microbiota disorders.

**Keywords:** oats; celiac disease; non-celiac gluten sensitivity; intestinal microbiota; gluten-free; SCFAs

### **1. Introduction**

Gluten-related disorders form an umbrella for all conditions related to gluten ingestion. These include, most importantly, celiac disease (CeD) and non-celiac gluten sensitivity (NCGS). The prevalence of these disorders has increased over the past 50 years, which makes them emerging health problems worldwide. Celiac disease is a chronic, systemic autoimmune disorder caused by gluten proteins in genetically susceptible individuals. In addition to CeD patients, NCGS subjects also require treatment with a gluten-free diet (GFD). These individuals develop adverse reactions such as gastrointestinal and extra-intestinal symptoms after exposure to gluten [1,2]. A life-long exclusion of gluten from the diet is currently the only effective treatment in alleviating the symptoms of these disorders. The adherence to a GFD and the following recovery from mucosal damage can be assumed to improve the nutritional status of the CeD patients observed at diagnosis. However, a long-term, strict GFD may be challenging to maintain due to social and economic burdens. Even when maintained,

GFD may be restricted and nutritionally suboptimal, since many gluten-free products have high fat and sugar but low fiber content. Such a diet predisposes patients to constipation, obesity and cardiovascular diseases [3–5].

The use of nutritious and fiber-rich whole-grain oats would diversify the GFD and improve the palatability, texture and fiber-content of the diet. Pure oats are being grown and produced following strict agricultural practices to minimize the contamination with other cereals. In Finland, oats are a major ingredient in the traditional daily diet and since the year 2000, pure oats have been considered suitable for the gluten-free diet [6]. Nowadays, oat products are widely used among Finnish celiac disease patients [7]. Although the inclusion of oats on GFD is recommended in Nordic countries, it is still not globally applied, possibly due to the debate regarding the safety of oats for CeD patients [8,9].

The intestinal microbiota primes the immune system and provides enzymes that expand the metabolic capacity of the host. The conversion of dietary components, such as dietary fiber, that escape the digestion of the host enzymes, support also the growth of microbes themselves. Intestinal microbiota and its metabolites play a major role in defining the antigen milieu of enterocytes, since they are able to interfere with the cells of the intestinal epithelium and modulate the signaling pathways through specific receptors [10]. It is assumed that a decreased microbiota diversity and relative abundances of specific bacterial taxa may lead to functional imbalance where the mutualistic relationship between the host and his microbes is disturbed. Indeed, deviations in the microbiota community structure have been associated with several local and systemic diseases, possibly contributing to the pathogenesis and/or clinical manifestation of these diseases (reviewed in [11]). In addition, GFD as such has been associated with potentially harmful alterations in microbiota, such as decreased microbiota richness, decreased amounts of bifidobacteria, lactobacilli as well as *Faecalibacterium prausnitzii* and increased amounts of *Proteobacteria* [12,13]. However, currently, the majority of the studies published on the fecal microbiota of celiac disease patients have been conducted with pediatric patients or by using conventional methods with limited throughput (reviewed in [14]).

To our understanding, the present study is one of the first on the gastrointestinal well-being and intestinal microbiota of persons with NCGS and within the few evaluating the intestinal microbiota of adult oat-using CeD subjects. The aim of this study was to evaluate the effect of oat consumption on the dietary status and gut well-being among adult subjects with gluten-related disorders who consume oat products on daily basis compared to healthy, oat consuming controls by using fecal microbiota signatures and its metabolites (short-chain fatty acids (SCFAs), ammoniacal nitrogen and β-glucuronidase activity) as biomarkers.

#### **2. Materials and Methods**

#### *2.1. Subjects and Study Design*

Celiac disease patients on a remission state (on a GFD at least 1 year), subjects with non-celiac gluten sensitivity (self-reported symptoms occurring after consuming a gluten-containing diet and adherence to a GFD for least 1 year) and healthy controls were recruited to the study. We decided not to test our NCGS subjects according to the Salerno criteria involving a separate gluten challenge trial for reasons discussed in later chapters [15]. The total number of the subjects recruited was 74, of which 49 completed the whole study period. After analyzing the food diary data, 6 subjects from NCGS group were excluded due to the consumption of gluten-containing food products. Thus, samples from celiac disease (CeD) patients (*n* = 19), non-celiac gluten sensitive subjects (*n* = 10) and healthy subjects (*n* = 14) were available for the further analyses. Based on food frequency questionnaire (FFQ) and 4d food diaries, all study subjects reported consumption of oat products daily. Demographic characteristics of study groups are presented in Table 1. Study subjects were recruited to the study from Turku region, Finland during the period August 2017–April 2018. Exclusion criteria were BMI below 18 or above 30, antibiotic treatment within the previous 6 months, use of any medication with gastrointestinal effects (e.g., laxatives or proton pump inhibitors) and blood donation or participation to another

clinical study within a month. Before the study entry, the volunteers were interviewed to assess the eligibility of the study. The subjects were ascertained to be in good health by means of self-reporting and normal results in screening blood tests (total blood count, fasting glucose and liver, kidney and thyroid functions, wheatspecific immunoglobulin E (IgE), total immunoglobulin A (IgA) and IgA antibodies to tissue transglutaminase (tTGAbA)). After the screening tests, the study subjects were enrolled in the study and were instructed to keep gut symptom diaries for 30 days. Study subjects consumed their habitual diet throughout the study period and were asked to fulfill food diaries during the last four days of the study. In addition, volunteers were asked to fulfill an FFQ of their dietary habits. Based on the FFQ, The Index of Diet Quality was calculated as explained in detail by Leppälä et al. [16] to assess the adherence to a health-promoting diet. Fecal samples for the microbiota, SCFAs, β-glucuronidase and ammoniacal nitrogen analyses were collected on the last day of the study period. The study protocol was approved by the Ethics Committee of the Hospital District of Southwest Finland (Identifier: ETMK:42/1801/2016) and subjects were enrolled in the study after written informed consent was obtained. The study was registered at ClinicalTrials.gov (Identifier: NCT02761785).



Dietary data are presented as an average of 4d intake based on food diaries. Values are mean (SD), unless otherwise stated. CeD subjects with celiac disease, NCGS non-celiac gluten sensitivity, CTRL healthy controls. <sup>1</sup> Pearson Chi-Square. Others One-way ANOVA. <sup>2</sup> median (min, max) Values with different letters differ from one another in each row.

#### *2.2. Dietary Intake Using Food Diaries*

Subjects were given written and oral instructions on filling the food diaries during the four days preceding the last study visit (including at least 1 weekend day). Kitchen scales were provided to ensure accuracy. Mean daily intakes of energy and macronutrients were calculated by using computerized software (AivoDiet 2.0.2.3; Aivo, Turku, Finland) utilizing the food composition database provided by the Finnish National Institute for Health and Welfare [17].

The quality of overall diet was assessed by FFQ validated for the evaluation of diet quality index [16]. The questionnaire contains 18 questions regarding the frequency and amount of consumption of food products during the preceding week. The quality of the diet was defined as poor when index points were less than 10 out of the maximum 15 points and good when points were 10 or more.

#### *2.3. Gut Symptom Diaries*

For the 30 day report of perceived gut symptoms, the study subjects were asked to mark down the type of the symptom (upper abdominal pain, lower abdominal pain, cramping, bloating, flatulence, bowel movement, diarrhea or constipation), the severity of the symptom in a scale of 1 to 3 (one meaning mild pain, two being moderate pain and three being intense pain), and the duration of the symptom. The diary was divided into time slots of three hours, except night time, which was marked as six hours slot (from midnight until 6 a.m.).

#### *2.4. Fecal Samples and DNA Extraction*

Fecal samples were frozen immediately after collection (20 ◦C) and stored at −70 ◦C once arrived in the research laboratory which was typically during the defecation day. Microbial DNA was extracted from fecal samples using the repeated bead—beating with KingFisher®—method as described in detailed previously [18]. The quality and quantity of the received DNA were measured by using a Nanodrop 1000 spectrophotometer (Thermo Scientific, Wilmington, DE). The quality of the DNA was good in all samples (OD 260/280 ratio ≥ 1.8).

#### *2.5. 16S Library Preparation*

The library preparation was started from 12.5 ng of total DNA. For NGS (Next-Generation Sequencing) library preparation, the recommended protocol for preparing 16S ribosomal RNA gene amplicons for the Illumina MiSeq system was used (Illumina 2013). The suggested universal bacterial primers were utilized for amplifying the V3 and V4 hypervariable regions of the bacterial 16S rRNA gene with polymerase chain reaction (PCR) using the KAPA Hifi HotStart Ready Mix (Roche Diagnostics Deutschland, Mannheim, Germany). PCR products were purified, and dual indices and Illumina sequencing adapters were attached using the Nextera XT index kit, Illumina. Finally, the libraries were purified once more with AMPure XP beads, Agencourt. The high quality of libraries was ensured using Advanced Analytical Fragment Analyzer and the concentrations of the libraries were quantified with Qubit® Fluorometric Quantitation (Life Technologies, Invitrogen division, Darmstadt, Germany). In a second PCR sample-specific "barcode"—primers and adapter sequences were attached. Up to 96 libraries were normalized and pooled for an Illumina MiSeq sequencing run using the MiSeq Reagent Kit version (v.) 3 with marginally overlapping 300 base pairs (bp) paired-end reads.

#### *2.6. 16S rDNA Sequencing*

The libraries were normalized and pooled for the automated cluster preparation, which was carried out by Illumina MiSeq instrument. Phix control library was added to the sequencing pool to balance the sequencing run. The libraries were sequenced in a single 2 × 300 bp run with Illumina MiSeq instrument using v3 sequencing chemistry. The sequencing run used paired-end sequencing chemistry with 8 bp dual index run.

#### *2.7. Short Chain Fatty Acids Assay*

The amounts of fecal short-chain fatty acids (SCFAs) were measured by solid-phase microextraction coupled to gas chromatography and mass spectrometry (SPME-GC-MS) to evaluate the microbial metabolic activity. Fecal samples (0.1 g) were weighted and suspended into 5 mL of deionized water by vortexing. 1.5 mL of fecal suspension was added into 10 mL vial with 0.5 g of NaH2PO4 [19]. Acetic acid, propanoic acid and butyric acid (Sigma-Aldrich, WGK Germany) were used as external standards in order to control the daily variation of instrument and sample preparation. The SPME fiber used was 75 μm CAR/PDMS, Fused Silica (Supelco, Bellefonte, PA, USA). The SPME-GC-MS analysis was carried out with Thermo Trace 1310—TSQ 8000 Evo equipped with an autosampler (Thermo Scientific, Wilmington, DE, USA). Compounds were separated by Supelco fused silica capillary column SPB-624, (30 m × 0.25 mm × 1.4 μm) under a carrier gas (helium) 1 mL/min with a splitless mode. The oven temperature program was as follows: 40 ◦C hold for 2 min and then 5 ◦C/min rise until 200 ◦C, hold for 10 min. A voltage of 70 eV was set in the EI. The system was operated using Xcalibur 4.0 (Thermo Scientific, Wilmington, DE, USA). Compounds were identified by the NIST library [20] and quantified by comparison to external standards. To optimize the SPME analysis, five commercial fibers were screened: 50/30 μm DVB/CAR/PDMS, 65 μm PDMS/DVB Stableflex, 65 μm PDMS/DVB Fused Silica, 100 μm PDMS and 75 μm CAR/PDMS. 75 μm CAR/PDMS was evaluated by comparison of SCFA standard runs as the most suitable for SCFA detection and chosen for the analysis.

#### *2.8.* β*-Glucuronidase and Ammoniacal Nitrogen Assays*

The activity of β-glucuronidase enzyme and the amount of ammoniacal nitrogen were measured from fecal samples to evaluate differences in these potentially harmful microbial metabolic activities. Ammoniacal nitrogen assay was carried out by an indophenol blue method reported in detail elsewhere [21]. Briefly, 0.1 g of wet fecal sample was diluted with 5 mL of deionized water, shaken for 60 min, and centrifuged at 3000× *g* for 3 min. Ammoniacal nitrogen concentration was measured from supernatant based on absorbance measured at 630 nm (Hidex Sense microplate reader, Hidex Oy, Turku, Finland). β-glucuronidase assay was carried out by the protocol of Shen [22]. Briefly, 0.1 g of wet fecal sample was diluted with 5 mL of deionized water and shaken for 60 min. 0.1 mL of diluted sample was added into Eppendorf tube® with 0.4 mL of 2 mM *p*-nitrophenyl-β-d-glucuronide solution (Sigma Aldrich, WGK Germany). Suspensions were incubated in anaerobic conditions at 37 ◦C for 60 min, followed by addition of 0.5 mL of 0.5 M NaOH. This suspension was centrifuged at 3200× *g* for 10 min and absorbance was measured on 405 nm (Hidex Sense microplate reader, Hidex Oy, Turku, Finland).

#### *2.9. Statistical and Data Analyses*

Statistical analyses of food diary and microbial metabolites data were carried out using IBM SPSS Statistics 25 software. Normal distribution of data was tested with Shapiro–Wilks test and ANOVA with contrast test were used to determine the statistical differences between study groups.

In the preprocessing of the MiSeq sequencing reads, the workflow proposed by [23] was adapted. In summary, the reads were trimmed from the left at 25 bp and 10 bp for the forward and reverse reads, respectively; and from right at 245 bp and 230 bp based on manual inspection of the read quality summaries. The sequence variant table was constructed from the reads with DADA2 [24] based on DADA2-formatted training FASTA files that were derived from the Ribosomal Database Project's Training Set 16 and the 11.5 release of the RDP database [25]. The chimeras were removed. The phylogenetic tree was constructed with the DECIPHER [26] and phangorn R packages. The preprocessed data were converted into a phyloseq R object [27], and aggregated to the genus level with the microbiome R package (function aggregate\_taxa). The full details are available in the source code that is openly deposited at Zenodo.

The Principal Coordinates Analysis (PCoA) was done for compositional data based on Bray–Curtis dissimilarity. Alpha diversity (Shannon index) was estimated with the *microbiome* [28] and *vegan* [29] R packages. For standard data manipulation and visualization, the *tidyverse* and *ggplot2* R packages were used, respectively. Beta diversity was done with PERMANOVA using the vegan R package and 999 permutations. The analyses were done with genus-level clr-transformed abundance tables unless otherwise mentioned. The group-level comparisons for individual genera were done with DESeq2.

#### **3. Results**

#### *3.1. Dietary Intake and the Quality of Diet*

Based on the food diary data (4 days), NCGS subjects received a higher proportion of their energy (E %) from fat and lower proportion (E %) from carbohydrates when compared to healthy controls (*p* = 0.025 and *p* = 0.045, respectively) (Table 1). Additionally, the gluten-sensitive subjects tended to get more energy than celiac disease patients when adjusted per body weight (kcal/kg of body weight, *p* = 0.09, data not shown). The mean intake of dietary fiber was at the lower end of the recommendation level in the three groups (Table 1). The dietary quality assessed by the validated index of diet quality questionnaire was considered good in most of the study subjects, average diet quality indices being higher than 10 in most of the study subjects (Table 1).

#### *3.2. Gut Symptom Diaries*

The highest amounts of gut symptoms per subject was reported by the NCGS groups subjects (61.4) when compared to CeD and healthy controls (39.1 and 19.7, respectively) (*p* = 0.045). In all study groups, the most often reported symptoms were flatulence, bloating and lower abdominal pain.

#### *3.3. Intestinal Microbiota Signatures*

The total microbiota profiles were comparable between CeD, NCGS and healthy controls (Figure 1). No statistically significant differences were observed in microbiota richness (Figure 2) or diversity (data not shown) between the study groups. Phylum-level microbial abundances were characterized by a high inter-individual variation and no statistically significant differences were observed between the study groups (Figure 3). However, the abundance of *Bifidobacterium* tended to be higher in the control group compared to CeD and NCGS (*p* = 0.067), (Figure 4).

#### *3.4. Microbial Metabolic Activity*

In CeD subjects, the amount of SCFAs was comparable to other groups (Table 2). However, the relative amount of acetate (% of total SCFAs) was higher in the control group compared to the NCGS group (*p* = 0.03). No statistically significant differences were observed in the proportions of propionate or butyrate between the groups. In addition, the amounts of ammoniacal nitrogen and β-glucuronidase activity were comparable between the study groups (Table 2).

**Figure 1.** Total microbiota profiles of study subjects were comparable between the groups as assessed by Principal Component Analysis (PCoA). CED celiac disease (*n* = 19), NCGS non-celiac gluten sensitivity (*n* = 10) and CTRL healthy controls (*n* = 14).

**Figure 2.** Microbiota richness was comparable in subjects with celiac disease (CED) (*n* = 19), non-celiac gluten sensitivity (NCGS) (*n* = 10) and healthy controls (CTRL) (*n* = 14). The box extends from 25th percentile to 75th percentile, with a line at median.

**Figure 3.** Relative abundances of bacterial phyla (% of total reads) in subjects with celiac disease (CED) (*n* = 19), healthy controls (CTRL) (*n* = 14) and subjects with non-celiac gluten sensitivity (NCGS) (*n* = 10). No statistically significant differences were observed between the study groups.

**Table 2.** Production of short-chain fatty acids (SCFAs), ammonia and the activity of β-glucuronidase in subjects with celiac disease (CeD), non-celiac gluten sensitivity (NCGS) and healthy controls (CTRL).


Concentrations are presented per g of fecal wet weight. Values are presented as mean (SD). Values with different letters differ from one another in each row.

**Figure 4.** The mean relative abundance of *Bifidobacterium* in subjects with celiac disease (CED) (*n* = 19), healthy controls (CTRL) (*n* = 14) and subjects with non-celiac gluten sensitivity (NCGS) (*n* = 10). The difference between the three was borderline significant (*p* = 0.067; Kruskal–Wallis test).

#### **4. Discussion**

Currently, the only treatment for celiac disease and other gluten-related disorders is a life-long adherence to a GFD. Due to the shortage of whole-grain products in the diet, GFD often results in inadequate intake of nutrients and dietary fiber [30,31] while in our study the average intake of dietary fiber intake was at the lower end of recommendation in all three groups, and did not differ between them. The difference may result from the fact that our subjects consumed oat products as a part of their habitual diet. Pure oats suitable for the gluten-free diet are grown, milled and handled without contamination by other cereals. Recently, it was suggested that the current confounding clinical findings on the safety of oat consumption in CeD subjects [32] could be caused by contaminated oats assessed as "pure" [8]. Indeed, a decade ago, gluten cross-contamination was shown to exist in oat supply chains in Europe, the United States and Canada [33]. Yet, our results suggest a great potential for pure oats as a source of fiber to GFD.

While persons diagnosed with CeD receive professional dietary advice in Finland (The Finnish Medical Society Duodecim 2018), many NCGS subjects are self-educated in GFD. In our study, the dietary composition of CeD subjects was comparable with healthy controls while subjects with NCGS obtained more energy from fat (>40 E %) and less energy from carbohydrates when compared to healthy controls. In addition, the food diaries in this study revealed that a large number of volunteers initially assigned to the NCGS group (6/16) consumed gluten, most often from rye or barley. This may be an indication that they are not aware of the composition of GFD. According to a recent survey study of Potter et al. [34], 24% of responded Australians avoided gluten completely or partially, while 14% had self-reported non celiac wheat sensitivity and 1% had celiac disease. Others avoided gluten for "general health" or as a treatment of abdominal pain, without being diagnosed with CeD or NCGS. The authors considered that gluten avoidance may be due to the current well-being trend and that NCGS may overlap with other gastrointestinal disorders [34]. Additionally, we observed that the NCGS group reported more gut symptoms per subjects when compared to CeD and healthy controls (*p* = 0.045). This finding is in line with a recent study by Tovoli et al. [35], where a significant proportion (66%) of gluten-sensitive subjects, diagnosed according to Salerno criteria, reported intestinal symptoms even years after the beginning of GFD. Compared to CeD patients following the same diet, subjects with NCGS reported a higher amount of symptoms (33%). Additionally, Skodje et al. (2019) reported a high number of gastrointestinal complaints among subjects with self-reported NCGS on a GFD.

The existence of NCGS as a condition has been recently challenged and intake of other non-gluten wheat components such as fructans [36] and amylase–trypsin inhibitors have been suggested to lie behind the symptoms instead of gluten as such [37]. Even a term change from NCGS to non-celiac wheat sensitivity has been suggested [38]. Still, NCGS has its defenders among consumers and researchers. The Salerno criteria [15] have been proposed for standardization of the diagnosis of NCGS. Criteria based investigation involves reporting of symptoms during 6 weeks on gluten-containing diet followed by 6 weeks of GFD and a further 1 week of test period containing GFD supplemented with either gluten test meals or placebo, 1-week washout and another test period in a cross-over manner. For the status of NCGS, 30% variation of symptoms between GFD and gluten containing diet periods is required. Such a gluten challenge was not imposed on the NCGS subjects in our trial due to limited resources and burden on the volunteers to participate on multiple clinical investigations. More so, our aim of this study was not to investigate which proportion of our self-reported NCGS volunteers would fulfil the Salerno criteria nor to limit our volunteers only to those getting gastrointestinal problems in the gluten challenge but to include subjects who self-reported their need for gluten-free diet despite lack of diagnosis for CeD or wheat allergy. Instead, the subjects were screened for negative celiac serology, specific immunoglobulin E (IgE) and wheat allergy. Generally, CeD patients are screened to ensure the remission state of their disease and NCGS patients to exclude the celiac disease before landing on NCGS diagnosis. However, once the GFD is initiated, testing for celiac disease is no longer accurate, which may lead to false-negative results in the case of self-diagnosed NCGS patients.

Previously, the majority of the studies analyzing the fecal microbiota of celiac disease patients have been conducted with pediatric patients or by using conventional methods with limited throughput [14]. The most often reported hallmarks of CeD microbiota have been increased abundances of Gram-negative bacteria, such as *Proteobacteria* and *Bacteroidetes* and reduced abundances of *Bifidobacterium* spp. and *Lactobacillus* spp. [14]. Similar changes have been also reported in studies examining the microbiota of healthy subjects after 1 month on GFD [12,13]. Moreover, some studies have reported persistent microbiota dysbiosis in CeD subjects in remission and on a GFD [39–41]. Of these studies [12,13,39–41], only Wacklin et al. (2014) report that the subjects consumed oats. Likewise, the studies reviewed by Marasco et al. (2016) concerning the microbiota composition of CeD patients or subjects following GFD, do not report oat consumption of the subjects, apart from the mentioned study of Wacklin et al. (2014). Our study on pure-oat consuming CeD subjects did not detect any signs of microbiota dysbiosis typically observed in CeD subjects with active disease nor detected any major GFD related changes in microbiota on NCGS or CeD subjects. The mean abundance of *Bifidobacterium* was higher in the control group compared to CeD and NCGS subjects but the difference was only marginally significant (*p* = 0.067). One of the CeD subjects had a high level of *Bifidobacterium* (9.7%). The abundance is within the typical range of variation for this genus, although it was unexpected to observe in the CeD group which has been associated with a reduced level of *Bifidobacterium*. This, combined with the moderate sample size of our study, may partially explain the only marginally significant difference between the groups. Therefore, our results do not support the hypothesis that a significant intestinal microbiota

dysbiosis would be the reason for the increased gastrointestinal symptoms reported by the NCGS group. The obtained results of microbiota composition agree with another Finnish study, where the CeD status of children who had consumed pure oats for 2 years was evaluated [42]. Small intestinal biopsies of these children showed normal histology and they had normal serological markers. Follow-up was continued for 7 years and all the markers remained normal during this period, suggesting that oats were well tolerated [42].

SCFAs are an important energy source for enterocytes and have been associated with several health-promoting effects including antipathogenic effects [43,44]. Their production varies among the individual microbiota compositions and by the type and amount of carbohydrates consumed [45]. We found that the relative amount of acetate (% of total SCFAs) was higher in the control group compared to NCGS group (*p* = 0.03), which may be linked to the higher abundance of *Bifidobacterium* in the control group compared to CeD and NCGS groups (*p* = 0.067) [46]. Proportions of propionate or butyrate did not differ between the groups. Thus, the intestinal microbiota of oat-using CeD and NCGS subjects was capable of producing similar amounts of propionate and butyrate than that of healthy controls. Previously, clinical studies assessing the SCFA levels have focused mainly on healthy adults [47–49], whereas adult celiac disease patients in remission have not been studied so far. Di Cagno et al. [50] analyzed volatile compounds from fecal samples of treated CeD children by SPME-GC-MS. The samples showed lower levels of SCFAs, such as butyric, isocaproic, and propanoic acids, when compared to healthy controls. However, the dietary habits of their subjects were not reported. It should be noted that SCFAs are rapidly absorbed in the colon and thus the fecal SCFA reflects losses rather than the amount of production in situ [43]. However, access to the proximal colon to quantify SCFA production rates is invasive and not possible in most study settings. Therefore, measurement of the fecal SCFAs is currently the only feasible way to estimate the production of compounds by gut microbiota. In this study, the interindividual variation of free-living human volunteers was large, but within the range observed previously [47–49]. The variation could possibly have been influenced by restrictions on the other parts of the diet than oats, but such were not applied in this study.

β-glucuronidase enzymes expressed by the intestinal microbiota mediate the reactivation of molecules important in human health and disease. For example, microbial glucuronidases regenerate toxic carcinogens whose increased activities in the GI tract have been associated with a higher incidence of gastrointestinal diseases such as colon cancer, Crohn's disease and colitis as well as to high-fat diets [51]. In rodent studies, high consumption of dietary fiber has been associated with decreased activity of β-glucuronidase [22,52]. In addition, a clinical crossover trial with 28 overweight male subjects demonstrated significantly decreased β-glucuronidase activity after consumption of wholegrain wheat and rye, when compared to low fiber control diet [53]. Our results show that CeD and NCGS subjects, who consume oat products on a daily basis, have similar β-glucuronidase activity levels than healthy controls.

No differences were seen in the ammonia production among the study groups. Ammoniacal nitrogen is a microbial end product produced by the deamination of amino acids. It is harmful to the host in high amounts, and previously the consumption of dietary fiber has been detected to decrease its production [53–55]. A study comparing the impact of diet on in vitro fermentation properties of whole grain flours and brans from corn, oats, rye and wheat reported significantly reduced ammonia concentrations after fermentation of oats and rye when compared to corn or wheat [54]. In general, ammonia concentrations of all groups studied were in line with previous results measured from healthy adults [47,56–60].

The strengths of this study were accurate analyses of fecal microbiota composition and metabolites of adult CeD and NCGS subjects, which has only received limited attention. A unique strength of the study was also the comparison of biological data to perceived symptoms by utilizing the gut symptom diaries. Moreover, the reliability of the results increased/enhanced the understanding of individual dietary habits of study subjects by analyzing 4d-food diaries as well as food frequency questionnaires assessing the overall quality of the study subject's diet. Additionally, the suitability of the study subjects was ensured by a detailed interview and screening at the recruitment to this study. The lack of non-oat-using CeD subjects can be considered as a limitation of this study since the inclusion of these subjects would have enabled a more accurate evaluation of the effect of the oat consumption on microbial biomarkers in celiac disease. However, the recruitment of non-oat using CeD subjects for the current study proved to be impossible. In Finland, the consumption of oats has been allowed for adult CeD subjects since 1997 and for pediatric patients since 2000 [6,61], and currently, most of the Finnish CeD subjects consume oats as part of their GFD [6,61].

To conclude, this study evaluated the influence of daily pure oat consumption on perceived and measured gut well-being in adult subjects with celiac disease and with subjects with non-celiac gluten sensitivity compared to healthy volunteers. No microbiota dysbiosis was detected among CeD nor NCGS subjects. However, further studies with metagenomic approaches should be conducted to assess the potential differences in microbiota composition and function in celiac patients consuming oats. The results of this study suggest that pure oats in a gluten-free diet represent a good alternative. We also demonstrate the need for further studies in NCGS subjects focusing on diet and microbiota interactions, and nutrition counseling to identify the causes of perceived gut symptoms.

**Author Contributions:** Conceptualization, K.M.L. and B.Y.; Methodology, L.N., S.H., S.S. and K.M.L.; Formal Analysis, L.N., S.H. and L.L.; Investigation, L.N., S.H., and L.L.; Resources, M.K., S.S., B.Y. and K.M.L.; Data Curation, L.N., S.H. and K.M.L.; Writing—Original Draft Preparation, L.N., S.H. and K.M.L.; Writing—Review & Editing, L.N., S.H., L.L., S.S., M.K., B.Y. and K.M.L.; Visualization, L.N., L.L., S.H. and K.M.L.; Supervision, K.M.L.; Project Administration, K.M.L.; Funding Acquisition, B.Y. and K.M.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was funded by Business Finland as part of the OATyourGUT project (grant number 5469/31/2016) co-funded by Finnish food companies and University of Turku, by Magnus Ehrnrooth Foundation (personal grant for SH) and by Academy of Finland (grant number 295741). The sequencing of the microbiological DNA was partially supported by Finnish Functional Genomics Centre, University of Turku, Åbo Akademi and Biocenter Finland.

**Acknowledgments:** All the volunteers who participated in this study are warmly acknowledged. Study nurse Sanna Himanen is acknowledged for the technical assistance in the screening blood tests. Annelie Damerau is thanked for technical advice in the SPME-GC-MS analyses. Annika Metsämarttila is acknowledged for discussions in the initial project planning phase.

**Conflicts of Interest:** The authors declare no conflict of interest.

**Data Availability Statement:** The cohort datasets generated and/or analyzed during the current study are not publicly available due to confidentiality, to protect the cohort participants' identity. The microbiota dataset analyzed in the current study is available from the corresponding author on reasonable request.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **Low Phytate Peas (***Pisum sativum* **L.) Improve Iron Status, Gut Microbiome, and Brush Border Membrane Functionality In Vivo (***Gallus gallus***)**

### **Tom Warkentin 1, Nikolai Kolba 2,**† **and Elad Tako 2,\*,**†


Received: 5 August 2020; Accepted: 20 August 2020; Published: 24 August 2020

**Abstract:** The inclusion of pulses in traditional wheat-based food products is increasing as the food industry and consumers are recognizing the nutritional benefits due to the high protein, antioxidant activity, and good source of dietary fiber of pulses. Iron deficiency is a significant global health challenge, affecting approximately 30% of the world's population. Dietary iron deficiency is the foremost cause of anemia, a condition that harms cognitive development and increases maternal and infant mortality. This study intended to demonstrate the potential efficacy of low-phytate biofortified pea varieties on dietary iron (Fe) bioavailability, as well as on intestinal microbiome, energetic status, and brush border membrane (BBM) functionality in vivo (*Gallus gallus*). We hypothesized that the low-phytate biofortified peas would significantly improve Fe bioavailability, BBM functionality, and the prevalence of beneficial bacterial populations. A six-week efficacy feeding (*n* = 12) was conducted to compare four low-phytate biofortified pea diets with control pea diet (CDC Bronco), as well as a no-pea diet. During the feeding trial, hemoglobin (Hb), body-Hb Fe, feed intake, and body weight were monitored. Upon the completion of the study, hepatic Fe and ferritin, pectoral glycogen, duodenal gene expression, and cecum bacterial population analyses were conducted. The results indicated that certain low-phytate pea varieties provided greater Fe bioavailability and moderately improved Fe status, while they also had significant effects on gut microbiota and duodenal brush border membrane functionality. Our findings provide further evidence that the low-phytate pea varieties appear to improve Fe physiological status and gut microbiota in vivo, and they highlight the likelihood that this strategy can further improve the efficacy and safety of the crop biofortification and mineral bioavailability approach.

**Keywords:** pea; phytate; iron; bioavailability; bio active compound; in vivo; *Gallus gallus*; brush border membrane; microbiome

#### **1. Introduction**

Micronutrient malnutrition affects more than half of the global population, primarily in developing regions [1,2]. Iron (Fe), zinc (Zn), and vitamin A deficiencies are prominent health constraints worldwide [3]. In low-income countries, plants are the significant source of food. In crude cereal and legume foods, the low bioavailability of Fe and Zn leads to metabolic disorders that are associated with these nutritional factors. Hence, increasing the nutritional value of such types of dietary ingredients will contribute to the nutritional status of the target population. Mineral, phosphorous, and phytate content is much higher in bran than whole grain [4–6].

Field pea (*Pisum sativum* L.) is a main pulse crop grown for human consumption as a source of protein, carbohydrates, minerals, and bioactive plant-origin bioactive compounds, contributing to better metabolic health. In 2014, the global production of peas was 11.2 million tons [7]. The main component of pea is starch, which includes two polymers of d-glucose: amylose and amylopectin [8,9]. Because of the alterations in physiochemical characteristics between pulses and cereal starches, starch from pulses can deliver some specific features to food systems as high gelation temperature, resistance to shear thinning, increased elasticity, and high concentration of resistant starch [10].

In addition, field peas include bioactive compounds such as oligosaccharides, polyphenols, and phytate [11]. Water-soluble carbohydrates in peas comprise mostly disaccharides and oligosaccharides. The raffinose group of oligosaccharides (RFOs) is the most targeted in pea research. These factors include galactose molecules (linked by α-d-1, 6-glycosidic bonds) attached to sucrose [12]. Humans lack the essential enzymes that are essential to break down these RFOs, and this results in these oligosaccharides being digested by intestinal bacterial populations via fermentation, leading to elevated short-chain fatty acid production [13]. Furthermore, a recent study indicated that intra-amniotic administration of raffinose upregulated the expression of brush border membrane (BBM) functional proteins, downregulated the expression of Fe-related proteins (indicating improvement of dietary iron bioavailability), and elevated villus surface area. Furthermore, raffinose increased the richness and composition of probiotic populations, and it reduced that of pathogenic bacterial species. Overall, raffinose improved microbial population, dietary Fe bioavailability, and BBM functionality in vivo [14].

The main phenolic compounds found in peas comprise condensed tannins, flavonoids, and phenolic acids [15]. These phenolic compounds are found specifically in the seed coat and are biosynthesized via the phenylpropanoid pathway, with condensed tannin molecules being responsible for the seed-coat coloring [16]. In dark-colored hulls, tannin and flavonoid compounds are the majority of phenolic compounds; however, in seeds with clear hulls, phenolic acids are the main compounds [17]. Polyphenols in the seed coat present antioxidant and anti-mutagenic activity, shielding the seed from oxidative stress [18]. In field conditions, these compounds also deliver chemical resistance against pathogens and insect pests during the growing process of the plant [19]. Polyphenols in peas appear mostly as insoluble or bound forms, covalently bonded to structural components of the cell wall such as cellulose, hemicellulose, lignin, and pectin [20,21]. The polyphenolic composition of peas is predominantly interesting with respect to metabolic health, given their alleged protective properties against oxidative stress [15,22]. According to Campos-Vega [11] and Rochfort [23], isoflavone polyphenols are linked with biological pathways in the lessening of osteoporosis and cardiovascular disease, the deterrence of cancer, and treating symptoms related to menopause. Phenolic compounds also display anti-nutritional effects, and related research showed a decrease in the bioavailability of proteins triggered by phenolic compounds [24]. Phytate functions as a storage for phosphate and minerals in seeds that can be recovered during germination process [25]. Phytate was recognized as an anti-nutrient due to its ability to chelate with multivalent ions, specifically Zn, Ca, and Fe, inhibiting the body's capability to absorb dietary minerals by limiting their bioavailability [24]. There is increasing interest in utilizing pulses in wheat-based products with blends [26]. The demand for gluten-free products led to investigation of the nutritional characteristics of baked products from pulses like chickpea and lentil [27], as well as peas [28]. The rheological properties of pea flour, including the gelation properties of starch, may be considered when exploring the potential application of pea flour in baked goods. Recent uses for pulses could increase the demand for pulses with specific nutritional and rheological properties, which will increase the need to investigate the components affecting the nutritional and functional properties of pulses. It was previously demonstrated that low-phytate pea lines had higher Fe bioavailability than regular or standard pea [29]; in addition, pea varieties which were low-phytate combined with relatively higher carotenoid concentration in some cases resulted in a further increase in Fe bioavailability in vitro [30].

Biofortified staple foods are an effective instrument through which to address micronutrient deficiencies worldwide, with emphasis on Fe and Zn, in numerous target populations [1,31–35]. The in vivo (*Gallus gallus*) model was established as an excellent model to assess dietary Fe and Zn bioavailability [33–39]. Hence, the objective of the current study was to evaluate the ability of low-phytate pea varieties in the context of a complete meal to improve Fe bioavailability and absorption, physiological status, intestinal BBM functionality, and intestinal microbial populations in vivo (*Gallus gallus*). We suggest the further use of in vivo screening model to guide future studies aimed to investigate biofortified staple food crops, as this method will allow proceeding to human efficacy studies with superior confidence and success.

#### **2. Materials and Methods**

#### *2.1. Plants Materials—University of Saskatchewan Pea Varieties*

The pea varieties evaluated in this research arose from the Crop Development Center, University of Saskatchewan (Canada) pea breeding program (Figure 1). Low-phytate line 1-2347-144 was derived from cultivar CDC Bronco [39,40] through chemical mutagenesis [41]. Varieties 4802-8-46Y-L, 4802-8-60G-L, and 4802-8-87Y-L resulted from the cross 1-2347-144/CDC 2235-4 made in 2011. CDC 2235-4 was later registered as CDC Raezer [42]. Variety 4803-4-78G-L resulted from the cross 1-150-81/CDC 2336-1 made in 2011. Line 1-150-81 is a second low-phytate line derived from CDC Bronco [41]. CDC 2336-1 was later registered as CDC Limerick [43]. The varieties from crosses 4802 and 4803 were previously described [30].

**Figure 1.** High-resolution photographs depicting six varieties used to evaluate the iron bioavailability of the Saskatchewan peas. To compare the differences in seed sizes, all photographs were taken to scale under standardized lighting conditions.

#### *2.2. Growing Conditions and Post-Harvest Handling*

All six pea varieties that were used in this experiment were grown at the Sutherland farm, located 10 km east of Saskatoon (Canada), with planting in May 2017 and harvest in August 2017. The harvested samples were stored in a non-heated warehouse, with temperature ranging between 15 and 20 ◦C based on the season, until shipment to Ithaca for dietary processing.

#### *2.3. Ingredient Preparation and Diet Composition*

For this study, raw pea seeds were rinsed and cleaned thoroughly in distilled water to remove dust, debris, and non-edible material. Peas were pre-soaked in distilled water (1:6 *w*/*w*) for 12 h at room temperature prior to cooking. Peas were cooked in boiling distilled water in stainless-steel steam kettles. Cooked peas were then stored at −20 ◦C for 24 h prior to freeze-drying (VirTis Research Equipment, Gardiner, NY, USA). Basmati rice and wheat were purchased from a local food store located in Ithaca, New York, USA. Our rationale with regard to the inclusion of basmati rice, wheat, and carrots in the tested pea-based diets was to approximately simulate the ingredients of a pea-based meal in India, which is one of the key consumers of pea, and where dietary Fe deficiency is a major health concern. Cooked rice was stored at −20 ◦C for 24 h before freeze-drying. Cooked/air-dried carrots were purchased from North Bay Trading Co. (Brule, WI, USA). Dried ingredients were milled into a course powder using a Waring Commercial® CB15 stainless-steel blender (Torrington, CT, USA). Other dietary ingredients included chicken Vitamin Mixture (#330002) and chicken Mineral Mix (#230000, no added iron) (Dyets Inc., Bethlehem, PA, USA), dl-methionine, and choline chloride (Sigma-Aldrich, St. Louis, MO, USA). The compositions of the experimental diets are shown in Table 1

#### *2.4. Iron Analysis*

Iron analysis was conducted as previously described [14,33,36,38,39]. For the analysis, a 500-mg sample of dietary ingredient, a 500-mg sample of pea-based diets, or a 100-mg sample of tissue (wet weight) was analyzed.

#### *2.5. Phytate Analysis*

Phytate (phytic acid) determination was conducted as previously described [14,33–39]. For the analysis, a 500-mg sample of dietary ingredients and a 500-mg of pea-based diets were analyzed, according to a phosphorous kit (K-PHYT; Megazyme International, Ireland).

#### *2.6. Protein and Fiber Analysis*

Analysis was conducted as previously described [36,43–45].

#### *2.7. Animals and Feeding Trial Design*

Cornish-cross fertile broiler eggs were delivered from a commercial hatchery (Moyer's Chicks, Quakertown, PA, USA). The eggs were incubated under ideal conditions at the Cornell University Animal Science poultry farm incubator. Upon hatch (hatchability = 98%), hatchlings were arbitrarily divided into seven treatment groups (*n* = 15) (Table 1), with ad libitum access to food and water (Fe concentration < 0.4 μg/L). Chicks were kept in a total confinement building (two animals per 1-m<sup>2</sup> metal cage) under controlled temperature and humidity with 16 h of light. Cages were equipped with an automatic watering system and a manual self-feeder. Feed intakes were documented daily, and, as of day of hatch, body weights were documented weekly. Animal protocols were approved by the Cornell University Institutional Animal Care and Use Committee (protocol number 2007-0129).

#### 2.7.1. Blood Collection, Hemoglobin, and Physiological Fe Status Parameters

Blood samples were collected and hemoglobin (Hb) assays were conducted according to the Hb kit manufacturer's instructions (BioAssay Systems, Hayward, CA, USA). Total body hemoglobin Fe (Hb-Fe), a parameter of iron absorption, was calculated from Hb concentrations and blood volume according to specific body weight (85 mL per kg of body weight) [33–36,39,46].

Hemoglobin maintenance efficiency (HME) was calculated as the cumulative difference in total body Hb Fe from the start of the study, divided by total dietary Fe intake. [33–36,39,46].

Upon the conclusion of the study (42 days), animals were euthanized by CO2 exposure and blood, small intestine, cecum, and liver samples were collected. Tissue samples were instantly frozen in liquid nitrogen and stored at −80 ◦C in a freezer until analyzed.

#### 2.7.2. Liver Iron and Ferritin

The quantifications of liver Fe and ferritin were conducted as previously described [46–48].

#### 2.7.3. Isolation of Total RNA from Duodenum

Total RNA extraction was conducted as previously described [14,33–39,46,49], according to the manufacturer's protocol (RNeasy Mini Kit, Qiagen Inc., Valencia, CA, USA).

#### 2.7.4. Real-Time Polymerase Chain Reaction (RT-PCR)

The complementary DNA (cDNA) reaction was conducted as previously described (BioRad C1000 touch thermocycler using the Improm-II Reverse Transcriptase Kit, Promega Corp., Madison, WI, USA) [37–39].



*Nutrients* **2020**, *12*, 2563

#### 2.7.5. Primer Design for Duodenal Gene Expression

Primers sequences were designed and selected using the Real-Time Primer Design Tool software (IDT DNA, Coralvilla, IA, USA). The *Gallus gallus* primers (forward/reverse) that were used in this study are indicated in Table 2.



<sup>1</sup> DMT-1, divalent metal transporter-1; DcytB, duodenal cytochrome b; ZnT1, zinc transporter 1; AP, amino peptidase; SGLT-1, sodium-glucose transporter-1; SI, sucrose isomaltase; 18S rRNA, 18S ribosomal RNA subunit.

#### 2.7.6. Real-Time qPCR Design

Isolated cDNA was used for the reaction (Cat. #1725274, Hercules, CA, USA) as previously indicated [36–39].

#### 2.7.7. Collection of Microbial Samples and DNA Isolation of Intestinal Contents

The cecum was removed and stored at −80◦C until analyzed. Microbial DNA isolation was conducted as previously described [36–38].

#### 2.7.8. Primer Design and PCR Amplification of Bacterial 16S rRNA

Primers for *Bifidobacterium, Lactobacillus, Escherichia coli*, and *Clostridium* were used in accordance with previously published data [46].

#### 2.7.9. Glycogen Analysis

At the conclusion of the study (day 42), the pectoral muscle (200 mg) was removed, and glycogen contents were determined as previously described [50–52].

#### *2.8. Statistical Analysis*

Statistical analyses were conducted using IBM SPSS Statistics 25 (IBM Analytics, Armonk, NY, USA). Measured parameters were found to have a normal distribution and equal variance, and they were acceptable for ANOVA. Mean separations for measured parameters were determined using ANOVA with the model including dietary treatment (seven levels) as the fixed effect, followed by a Duncan post hoc test. Differences with *p*-values ≤0.05 were considered statistically significant.

#### **3. Results**

#### *3.1. Seed Iron and Phytate Concentrations in Experimental Peas Varieties*

Iron concentrations of dietary ingredients are shown in Table 1. Differences in seed Fe contents in the pea varieties were significant (*p* ≤ 0.05), ranging from 37 μg/g in 1-2347-144 to 42 μg/g in 4803-4-78G-L (Table 1). Phytate concentrations and molar ratios of dietary ingredients of the pea-based diets are indicated in Table 1. Significant (*p* ≤ 0.05) differences in phytate concentrations were measured between peas varieties, from 3.7 mg/g in 4803-4-78G-L to 5.82 mg/g in CDC Bronco (Table 1). Phytate-to-Fe molar ratios varied significantly (*p* ≤ 0.05), from a ratio of 7.4 in 4803-4-78G-L to a ratio of 12.4 in CDC Bronco (Table 1).

#### *3.2. Protein and Fiber Contents*

Table 3 indicates the total crude protein content in experimental tested pea varieties, with significant differences (*p* ≤ 0.05) between pea varieties, ranging from 22.5 g/100 g in CDC Bronco to 26.75 g/100 g in 4803-4-78G-L. Concentrations of insoluble, soluble, and total fiber for experimental peas are shown in Table 3, with significant differences (*p* ≤ 0.05) in each of the fiber fractions between experimental peas. The lowest concentrations of the insoluble, soluble, and total fiber were detected in the 4803-4-78G-L pea variety. Significantly (*p* ≤ 0.05) higher concentrations of all three fiber fractions were measured in 1-2347-144. As a reference, the total protein content in the control diet (no pea) was measured at 10.72 g/100 g ± 0.16 g/100 g of total protein.


**Table 3.** Protein and fiber concentrations (g/100 g) of tested peas varieties 1.

<sup>1</sup> Values are means ± standard error of the mean (SEM) (*<sup>n</sup>* <sup>=</sup> 3 replicates). a–d Treatment groups not indicated by the same letter are significantly different (*p* ≤ 0.05).

#### *3.3. Iron–Phytate Analysis of Pea Based Diets*

The final composition of the six pea-based diets and no-pea diet are shown in Table 3. Iron concentrations amongst the pea-based diets were significantly different (*p* ≤ 0.05). Diets formulated from 4802-8-87Y-L and 4803-4-78G-L had the highest iron concentrations (38 μg/g and 39 μg/g, respectively) relative to the control diet (no-pea diet) (27 μg/g). Final phytate concentrations also varied between experimental diets ranging from 1.57 mg/g in 1-2347-144 to 2.66 mg/g in the no-pea diet. Significant (*p* ≤ 0.05) differences in phytate–Fe molar ratios were observed between the pea-based diets, ranging from 3.79 mg/g in 1-2347-144 to 8.66 mg/g in CDC Bronco (Table 1).

#### *3.4. In Vivo Assay (Gallus gallus Feeding Trial)*

3.4.1. Growth Rates, Hemoglobin (Hb), Total Body Hemoglobin Fe (Hb-Fe), and Hemoglobin Maintenance Efficiency (HME)

Feed intakes and Fe intakes were higher (*p* < 0.05) in all pea-based dietary treatment groups relative to the no-pea dietary treatment group (Tables 4 and 5).


**Table 4.** Experimental cumulative feed intake 1.

<sup>1</sup> Values are means ± SEM (*n* = 15 animals per treatment group). a,b Treatment groups not indicated by the same letter are significantly different (*p* ≤ 0.05).


**Table 5.** Experimental cumulative iron intake 1.

<sup>1</sup> Values are means ± SEM (*n* = 15 animals per treatment group). a–c Treatment groups not indicated by the same letter are significantly different (*p* ≤ 0.05).

Also, as from day 35 of the study, body weights were consistently higher (*p* < 0.05) in several of the low phytate pea based dietary groups (4803-4-78G-L, and 4802-8-87Y-L), relative to the CDC Bronco and no-pea dietary groups (Table 6). Hemoglobin (Hb) values did not differ between treatment groups; however, significant differences in total body Hb-Fe, a physiological biomarker of Fe bioavailability and status, were detected as of week five of the study (Table 7), demonstrating an improvement in Fe status in the 4802-8-87Y-L group, relative to CDC Bronco and no-pea diet groups. In addition, the standard pea variety treatment group (CDC Bronco) had a lower HME (*p* < 0.05) at each time point when compared to the group receiving the lower-phytate pea-based diets (groups 1-2347-144, 4803-4-78G-L), indicating a higher dietary Fe bioavailability and increased absorbable Fe (Table 8).

**Table 6.** Experimental body weights 1.


<sup>1</sup> Values are means ± SEM (*n* = 15 animals per treatment group). a–d Treatment groups not indicated by the same letter are significantly different (*p* ≤ 0.05). Body weights averaged 38 g at the start of the experiment.

**Table 7.** Experimental total body hemoglobin iron (Hb-Fe) 1.


<sup>1</sup> Values are means ± SEM (*n* = 15 animals per treatment group). a–d Treatment groups not indicated by the same letter are significantly different (*p* ≤ 0.05). Total body hemoglobin iron averaged 0.65 milligrams at the start of the experiment.


**Table 8.** Experimental hemoglobin maintenance efficacy (HME) 1.

<sup>1</sup> Values are means ± SEM (*n* = 15 animals per treatment group). a–c Treatment groups not indicated by the same letter are significantly different (*p* ≤ 0.05).

#### 3.4.2. Hepatic Iron and Ferritin Concentrations

The contents of liver iron and ferritin (day 42) are shown in Table 9. Significant (*p* ≤ 0.05) differences in liver iron were detected among the seven treatment groups with concentrations ranging from 73 μg/g in the group receiving the 4803-4-78G-L diet to 96 μg/g in the 1-2347-144 diet. Significant (*p* ≤ 0.05) differences in liver ferritin concentrations were also measured between the seven dietary treatment groups (Table 9).

**Table 9.** Hepatic iron and ferritin protein concentrations 1.


<sup>1</sup> Values are means ± SEM (*n* = 12 animals per treatment group). a–c Treatment groups not indicated by the same letter are significantly different (*p* ≤ 0.05). Total iron concentrations were measured as micrograms per gram of liver tissue (wet weight). Liver ferritin concentrations were measured as arbitrary units of liver tissue (wet weight).

#### 3.4.3. Serum Iron Concentrations

Significant differences (*p* ≤ 0.05) in serum iron concentrations were detected on day 21 and 35 of the study. On day 21, the lowest concentration of serum iron was 1.526 μg/μL in the no-pea dietary group, while the highest concentration was in the 4802-8-87Y-L pea-based dietary group (2.812 μg/μL). On day 35, the lowest concentration of serum iron was 1.488 μg/μL (no-pea dietary group), while the highest concentration was detected in the 4803-4-78G-L dietary group (2.633 μg/μL) (Table 10).

**Table 10.** Serum iron concentrations 1.


<sup>1</sup> Values are means ± SEM (*n* = 12 animals per treatment group). a,b Treatment groups not indicated by the same letter are significantly different (*p* ≤ 0.05).

#### 3.4.4. Glycogen Concentrations in Pectoral Muscle

As an indicator of energetic status [52,53], pectoral muscle glycogen concentrations were measured on days 21 and 42 of the study (Table 11). No significant differences were detected on day 21; however, significant differences (*p* ≤ 0.05) were measured on day 42 in the abundance of glycogen stored in pectoral muscles. The highest values of glycogen were in the 4802-8-60G-L pea-based dietary group, and the lowest concentration of glycogen was in the no-pea dietary group.


**Table 11.** Pectoral muscle glycogen concentrations (AU) 1.

<sup>1</sup> Values are means ± SEM (*<sup>n</sup>* <sup>=</sup> 5 animals per treatment group. a,b Treatment groups not indicated by the same letter are significantly different (*p* ≤ 0.05).Glycogen concentrations were measured as milligrams per milliliter of pectoral tissue (wet weight).

#### 3.4.5. Duodenal Gene Expression

The duodenal gene expression of iron- and zinc-related proteins, as well as BBM functional proteins, is shown in Figure 2. Significant (*p* ≤ 0.05) differences in the expression of DcytB and ferroportin were identified, with no significant differences in divalent metal transporter-1 (DMT1) expression between treatment groups.

**Figure 2.** Gene expression of iron proteins in the duodenum after six weeks of consuming pea-based diets. Values are means <sup>±</sup> SEM (*<sup>n</sup>* <sup>=</sup> 10 per treatment group). a–c Treatment groups not indicated by the same letter are significantly different (*p* < 0.05). DMT-1, divalent metal transporter-1; DcytB, duodenal cytochrome b; ZnT1, zinc transporter 1; AP, amino peptidase; SGLT-1, sodium-glucose transporter 1; SI, sucrose isomaltase.

#### 3.4.6. Cecum Content Bacterial Populations Analysis

As shown in Figure 3, the relative abundance of *Bifidobacterium* was significantly higher (*p* < 0.05) in the 4802-8-87Y-L and CDC Bronco groups relative to all other treatment groups. Furthermore, the abundance of *Lactobacillus* was significantly higher (*p* < 0.05) in the 1-2347-144 and 4803-4-78G-L groups relative to all other treatment groups.

**Figure 3.** Genus- and species-level bacterial populations (AU) from cecal contents after six weeks of consuming pea-based diets. Values are means <sup>±</sup> SEM (*n* = 10 per treatment group). a–c Treatment groups not indicated by the same letter are significantly different (*p* < 0.05).

#### **4. Discussion**

The objective of the current study was to investigate the effects of low-phytate peas, in the context of a complete meal, on Fe bioavailability, absorption, physiological status, intestinal BBM functionality, and gastrointestinal microbial populations in vivo (*Gallus gallus*).

In studies of biofortification, the process via which the nutritional quality of food crops is improved through agronomic practices, conventional plant breeding, or modern biotechnology [2], it is necessary and advantageous to utilize in vivo screening tools that are capable of assessing biofortified varieties of staple crops, as well as in relation to the diet in which they are consumed [1,33,36,38,39,46,54–56]. The present study, for the first time, presents a demonstration of how the *Gallus gallus* model of Fe (and Zn) bioavailability could be useful in the design of the current study aimed at assessing the potential nutritional benefit of lower-phytate versus standard peas. The chosen dietary composition was specifically formulated in accordance to a potential target population (Indian/Bangladeshi pea-based dal meal), similar to previous in vivo studies aimed at assessing dietary Fe bioavailability in beans [35,55] and wheat [38] (Table 1). Overall, our data agree with previously published knowledge [1,39,49,55], demonstrating that this in vivo screening approach is effective in the evaluation process of nutritional qualities of the low-phytate pea varieties. Furthermore, the data suggested that lower-phytate pea-based diets were able to moderately improve Fe physiological status in vivo.

Peas are a common staple food crop consumed worldwide, primarily in India, China, Russia, Ethiopia, and Bangladesh. Global dry pea production increased from 9.9 million tons in 2012 to 16.2 million tons in 2017 [7]. Currently, the leading producers are Canada, Russia, China, Ukraine, and India. In Canada, a leading producer and exporter of dry peas, pea was grown on 1.6 million ha in western Canada (Saskatchewan, Alberta, and Manitoba) in 2017, indicating a significant alteration in cropping practices from the 300 ha reported in 1967. Pea was the major alternative crop as farmers shifted toward a more diversified crop production. Pea varieties (yellow and green cotyledon) are grown, with an average of 80% production in yellow cotyledon varieties. The five-year (2013–2017) average pea yield in western Canada is 2.6 tons/ha (38 bu/ac) [57]. As for their nutritional value, it was previously demonstrated that pea seeds are high in protein, carbohydrates, fiber, B vitamins, and minerals (potassium, magnesium, calcium, iron), and they are considered an inexpensive source of energy-dense, nutrient-rich food [58–60]. In addition, pea seeds are low in fat and cholesterol-free. Because of these nutritional benefits, worldwide pea utilization is expected to continue to grow.

Plant seeds, such as pea, contain a high concentration of phosphorus. However, about 60–80% of the total phosphorus in seeds is stored in the form of phytate, a mixed-cation salt of phytic acid [59]. This introduces a nutritional challenge, as negatively charged sites of phytic acid bind and form salts with K<sup>+</sup>, Mg2<sup>+</sup>, Ca2<sup>+</sup>, Mn2<sup>+</sup>, Zn2<sup>+</sup>, or Fe3<sup>+</sup> [61]. Phytate causes multiple difficulties, as non-ruminant animals including pig, poultry, fish, and humans, are unable to digest phytate due to lack of a phytase enzyme [61]; as a result, important micronutrients (as Zn2<sup>+</sup> and Fe3<sup>+</sup>) bound to phytate are also excreted and not absorbed, potentially leading to micronutrient deficiencies [62]. Recently, the development of cultivars with low-phytate content became an effective approach to potentially reducing nutritional concerns ascending from the consumption of phytate-rich grains. Low-phytate varieties were chemically persuaded in maize (*Zea mays* L.) [63], soybean (*Glycine max* (L.) Merr.) [64], barley (*Hordeum vulgare* L.) [63,65], rice (*Oryza sativa* L.) [66], wheat (*Triticum aestivum* L.) [67], bean (*Phaseolus vulgaris* L.) [68], and pea [41]. The concentration of phytate phosphorus is significantly reduced in the mutants with an associated increase in available phosphorus. Wilcox et al. [65] reported an 80% reduction in phytate phosphorus content in a low-phytate soybean mutant, as compared with its nonmutant sibling, and this reduction was matched by an equal increase in inorganic phosphorus.

It was previously demonstrated that low-phytate crops increase the bioavailability of phosphorus and several important nutritional cations, including Fe. These crops could assist in increasing the health of a large proportion of the global population, which is dietary Fe-deficient, primarily in target regions where dietary peas are consumed regularly. For example, in a previous study focused on the nutritional evaluation of low-phytate pea diets in vivo, it was demonstrated that animals fed the low-phosphorus diets had lower weight gain and feed intake (*p* < 0.01) than those fed the higher phosphorus level. Bone strength was higher (*p* < 0.01) for animals fed diets based on low-phytate pea than for those fed diets based on normal pea or soybean meal. The authors concluded that increasing the availability of the phosphorus in peas could mean that less inorganic phosphorus would be required in order to meet the nutritional requirements of broilers [59].

In the context of the current study, the results indicated that, despite Hb levels not being significantly higher in the lower-phytate pea groups, significant differences in total body Hb-Fe, the physiological Fe status biomarker [33–36,39,46,55], were observed (Table 8), representing an enhancement in Fe status in the 4802-8-87Y-L dietary group, relative to CDC Bronco and the no-pea dietary group. In addition, the standard pea variety (CDC Bronco) treatment group had a lower HME (*p* < 0.05) ratio compared to the group receiving the lower-phytate pea-based diets (groups 1-2347-144, 4803-4-78G-L) (Table 8), indicating improved dietary Fe bioavailability and increased absorbable Fe [36,46,54]. The CDC Bronco diet presented a higher PA–Fe ratio compared to the all low-phytate pea-based diets (Table 1), which was associated with increased dietary Fe bioavailability in these pea-based diets [69–71]. These results agree with preceding experiments intended to assess Fe bioavailability in Fe-biofortified legumes, such as black beans [72], red mottled beans [33], Carioca beans [36], and pearl millet [73], as well as in the context of a complete diet. Thus, several intrinsic factors, including phytates, may influence the bioavailability of Fe from these pea varieties and other crops [56,74–76], potentially limiting their nutritional benefit.

Previous research suggested that increased Fe content alone in biofortified crops may not be adequate to produce a significant physiological improvement in Fe status and in Fe-deficient populations [36,55,76]. In the current study, it appears that, although Fe contents of all tested pea varieties were similar, the consumption of lower-phytate peas was able to moderately improve Fe status and storage, as further suggested by the hepatic ferritin contents of lower-phytate groups relative to CDC Bronco and no-pea diets. Furthermore, the duodenal brush border membrane (BBM) gene expression of ferroportin (FPN) was significantly upregulated, while DcytB was downregulated in the groups receiving the lower-phytate pea-based diets, relative to the CDC Bronco dietary group (*p* < 0.05, Figure 2). However, no significant alterations in the expression of BBM functional proteins were detected amongst treatment groups. Previous studies showed a downregulation of the gene expression of Fe-related BBM proteins (DMT-1, FPN, and Dcytb) in Fe-biofortified diets compared to the Fe-standard diets [36,46,55]. Ferroportin is the Fe exporter that transfers Fe across the enterocyte's basolateral membrane [77]. Hence, since the lower-phytate pea-based dietary groups had a higher expression of FPN, more Fe could be transported from the enterocyte into the blood and target tissue; therefore, this mechanism indicates the potential increased amount of absorbable Fe and, hence, the total body Hb-Fe increased in some of the low phytate groups compared to the CDC Bronco and no-pea dietary groups.

Similar to humans and most animals, the *Gallus gallus* model harbors a complex and active intestinal microbiota [78], significantly and directly influenced by host genetics, environment, and diet [79]. There is a significant resemblance at the phylum level between the gut microbiota of *Gallus gallus* and humans, with Bacteroidetes, Firmicutes, Proteobacteria, and Actinobacteria representing the dominant bacterial phyla in both [80]. In this study, a genus- and species-level bacterial population delineation among the low-phytate, standard (CDC Bronco), and no-pea dietary groups was observed. Results indicated that the abundance of *Bifidobacterium* was significantly higher (*p* < 0.05) in the 4802-8-87Y-L and CDC Bronco groups relative to all other treatment groups. Furthermore, the abundance of *Lactobacillus* was significantly higher (*p* < 0.05) in the 1-2347-144 and 4803-4-78G-L treatment groups relative to all other treatment groups (Figure 3). These results suggest that the above lower-phytate pea-based diets may potentially improve the host overall gut health by promoting the abundance of beneficial bacterial populations. Moreover, some of the low-phytate pea varieties (as 1-2347-144) presented a higher (*p* < 0.05) total fiber content (soluble and insoluble) compared to the standard CDC Bronco pea (Table 3). It was previously demonstrated that soluble fiber can increase villi height by elevating intestinal cell proliferation [81]. In the current study, some of the low-phytate pea dietary groups (such as 4803-4-78G-L, 4802-8-46Y-L, and 4802-8-87Y-L) presented higher (*p* < 0.05) protein content compared to the standard CDC Bronco pea (Table 3), where a higher dietary protein content was shown to increase villi height and intestinal cell proliferation [82]. Furthermore, indigested dietary proteins and fibers are fermented in the lower intestine, and this action produces short-chain fatty acids (SCFAs), such as acetate, propionate, and butyrate. Production of SCFAs affects metabolism and gastrointestinal health [83]. Acetate and propionate are energy substrates for peripheral tissues, and butyrate is referentially used as an energy source by colonocytes [84,85].

In summary, the current study focused on the performance of low-phytate pea varieties in chicken diets. Phytate phosphorus concentration was reduced by approximately 40% in these varieties. The low-phytate pea variety-based diets were able to moderately improve the Fe status in vivo, suggesting that low-phytate field pea has the potential to improve Fe bioavailability in human diets, particularly in the Indian subcontinent, as one of the major importing regions for Canadian peas, and a region where dietary Fe deficiency is a major health concern. Furthermore, as the abolition of micronutrient malnutrition remains a widespread global health problem in developing countries, the current study suggests that increasing micronutrient intake in food through food-based approaches is a sustainable method for the potential prevention of micronutrient deficiencies. Biofortification offers a long-term, sustainable, food-based solution for a world population, and breeding programs may aim to improve grain Zn and Fe concentrations; however, as previously suggested, improving Fe or Zn content may not necessarily result in the desired outcome (i.e., breeding toward increased mineral content may also lead to increased potential dietary inhibitors) and, hence, may not be as effective. In low-income countries, breeding for mineral solidity may remain the only agricultural involvement available to improve the nutritional content of staple crops, and, as suggested in the current study, the genetic improvement of staple food crops, specifically the development of low-phytate pea verities, resulted in improved nutritional quality and dietary Fe bioavailability, including in a complete diet context.

Additionally, as previously demonstrated, the current study presents a cost-effective approach designed to assess the effectiveness of biofortified pea varieties in vivo, as these varieties were developed with an aim to reduce the inhibitory effect of dietary phytate on Fe bioavailability. Therefore, our findings suggest that the use of lower-phytate biofortified peas may be an effective and sustainable approach to decreasing the global abundance of Fe deficiency, with added improvements in intestinal bacterial population structure and intestinal BBM functionality.

#### **5. Conclusions**

Nutritional approaches aimed to ease global Fe deficiency, such as Fe supplementation or fortification, are moderately successful at achieving optimal Fe status. This study showed how biofortified low-phytate pea affects dietary Fe bioavailability, physiological status, and the composition and metagenome of the gut microbiota and intestinal function. Animals (*Gallus gallus*) that consumed the low-phytate pea-based diets had increased abundance of beneficial bacteria, with associated surges in SCFA-producing bacteria with known phenolic catabolic capability, which resulted in an improvement in intestinal functionality. In addition, some of the low-phytate peas presented a higher protein content versus the standard CDC Bronco pea, which can possibly improve Fe bioavailability and intestinal functionality. Furthermore, parallel to preceding data, the current research suggests that a key aspect to include is the in vivo measurement of dietary Fe bioavailability in biofortified crop variety-based diets, as part of the plant breeding procedure.

Overall, our discoveries provide further evidence that, unlike other nutritional approaches to improving Fe status, the low-phytate pea varieties appear to improve Fe physiological status and gut microbiota in vivo, and they present an option for this strategy to further advance the efficacy and safety of crop biofortification and mineral bioavailability. We recommend the application of in vivo screening tools to guide studies aimed at developing and appraising Fe bioavailability in biofortified food crops, as well as their possible nutritional benefit. Based on the data presented in the current study, a human efficacy study will be conducted to compare the 4802-8-87YL (low phytate) and CDC Bronco (standard/normal phytate) varieties, along with a no-pea control.

**Author Contributions:** Data curation, T.W. and E.T.; formal analysis, N.K. and E.T.; investigation, T.W. and E.T.; methodology, N.K., T.W., and E.T.; resources, T.W. and E.T.; supervision, E.T.; writing—original draft, E.T.; writing—review and editing, T.W. and E.T. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors wish to thank Martino and Silva (Department of Nutrition and Health, Federal University of Viçosa, Viçosa, Minas Gerais, Brazil) for conducting the protein and fiber analyses.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
