Almond Consumption and Processing Affects the Composition of the Gastrointestinal Microbiota of Healthy Adult Men and Women: A Randomized Controlled Trial

Background: Almond processing has been shown to differentially impact metabolizable energy; however, the effect of food form on the gastrointestinal microbiota is under-investigated. Objective: We aimed to assess the interrelationship of almond consumption and processing on the gastrointestinal microbiota. Design: A controlled-feeding, randomized, five-period, crossover study with washouts between diet periods was conducted in healthy adults (n = 18). Treatments included: (1) zero servings/day of almonds (control); (2) 1.5 servings (42 g)/day of whole almonds; (3) 1.5 servings/day of whole, roasted almonds; (4) 1.5 servings/day of roasted, chopped almonds; and (5) 1.5 servings/day of almond butter. Fecal samples were collected at the end of each three-week diet period. Results: Almond consumption increased the relative abundances of Lachnospira, Roseburia, and Dialister (p ≤ 0.05). Comparisons between control and the four almond treatments revealed that chopped almonds increased Lachnospira, Roseburia, and Oscillospira compared to control (p < 0.05), while whole almonds increased Dialister compared to control (p = 0.007). There were no differences between almond butter and control. Conclusions: These results reveal that almond consumption induced changes in the microbial community composition of the human gastrointestinal microbiota. Furthermore, the degree of almond processing (e.g., roasting, chopping, and grinding into butter) differentially impacted the relative abundances of bacterial genera.


Introduction
Diet is a contributing factor to the development of numerous diseases including obesity, cardiovascular disease, and type 2 diabetes. Increasingly, these complex metabolic diseases are also being associated with the gastrointestinal microbiota [1]. The gastrointestinal microbiota, an ecosystem containing trillions of microorganisms, including bacteria, archaea, and fungi, is, in essence, a microbial organ that possesses more than 150 times more genes than the human genome [2]. Importantly, these microbes possess genes that allow for the metabolism of foods that escape digestion by human alimentary enzymes, thereby aligning diet as an important modulator of the interrelationship between diet, the gastrointestinal microbiota, and host health.
Diet is a key modulator of the human gastrointestinal microbiota-habitual dietary patterns [3], rapid changes in dietary macronutrient composition [4], eating behaviors [5], and increased intake of the gastrointestinal microbiota were assessed. Briefly, this was a controlled-feeding, randomized, cross-over study with five three-week diet periods, each diet period was separated by a one-week non-controlled diet break (wash-out) period. The controlled diets, fed at weight maintenance, consisted of a base diet designed to represent the typical American diet, with a macronutrient composition of 53% of kcals from carbohydrates, 32% of kcals from fat, and 15% of kcals from protein, and 0 g/day of almonds. All foods were identical between the control and almond diets, except for the almonds. During the four almond diet periods of the study, the controlled diet was scaled down to allow for the isocaloric inclusion of 42 g/day of varying forms of almonds: whole, natural almonds; whole, dry roasted almonds; chopped, dry roasted almonds; and dry roasted, almond butter.
Study participants were recruited in the Washington, DC metropolitan area surrounding the Beltsville Human Nutrition Research Center (BHNRC) in Beltsville, MD, USA. All subjects gave their informed consent for inclusion before participating in the study. The study was conducted in accordance with the Declaration of Helsinki, and the protocol and informed consent document were reviewed and approved by the Medstar Health Research Institute (Protocol #2013-136). Participants were excluded if they were <25 or >75 years of age, had a body mass index (BMI) <20 or >38 kg/m 2 , blood pressure ≥160/100 mm Hg, fasting blood glucose ≥126 mg/dL, fasting total blood cholesterol ≥280 mg/dL, and fasting triglycerides ≥300 mg/dL. Participants were excluded if they had any of the following diseases or conditions: allergy to almonds or other nuts, gastrointestinal disease, malabsorptive syndromes, liver disease, pancreatic disease, kidney disease, gout, hyperthyroidism, untreated or unstable hypothyroidism, certain cancers, or type II diabetes; history of bariatric or certain other surgeries related to weight control; history of eating disorders or other dietary patterns which are not consistent with the dietary intervention (e.g., vegetarians, very low fat diets, high protein diets); greater than 10% change in body weight within the last 12 months; pregnant women or women who plan on becoming pregnant during the study period; women who had given birth within the past 12 months; and tobacco users. A random number generator was utilized for study participant randomization following the creation of a randomization scheme that stratified the 18 participants (10 males; eight females) by sex and BMI. Each participant was assigned to a treatment sequence that consisted of each of the five diet periods.

Sample Collection and Analysis
Fecal samples were collected at the end of each of the five diet treatment periods (e.g., control, whole almonds, whole roasted almonds, roasted chopped almonds, and almond butter) for microbiota analyses. Following fecal collection and homogenization, samples were stored at −80 • C at the BHNRC prior to overnight shipment on dry ice to the University of Illinois and subsequent storage of samples at −80 • C until analysis. Fecal DNA was extracted according to manufacturer's guidelines using the MoBio Powerlyzer Powersoil Kit (MoBIO Laboratories Inc., Carlsbad, CA, USA). After extraction, sequencing was performed at the W. M. Keck Center for Biotechnology at the University of Illinois as previously described [24]. Briefly, bacterial (V4 region of 16S rRNA; 515F/806R: 5 -GTGCCAGCMGCCGCGGTAA, 5 -GGACTACHVGGGTWTCTAAT), archaeal (ArchaeaF/ArchaeaR: 5 -GYGCASCAGKCGMGAAW, 5 -GGACTACVSGGGTATCTAAT) and fungal (ITS1F/ITS4R: 5 -TTCGTAGGTGAACCTGCGG, 5 -TCCTCCGCTTATTGATATGC) genes were amplified using a Fluidigm access array (Fluidigm, South San Francisco, CA, USA). Following amplification, a fragment analyzer (Advanced Analytics, Ames, IA, USA) was utilized to check the quality of the amplicon pool and confirm amplicon region and size. Next, a DNA pool was generated by combining equimolar amounts of the amplicon from each sample. Then, amplicon pools were size-selected on a 2% agarose E-gel (Life Technologics, Grand Island, NY, USA) to remove low molecular weight DNA fragments. A Qiagen gel purification kit (Qiagen, Valencia, CA, USA) was used to extract the amplicon pools from the gel and remove impurities. Then, size-selected, cleaned amplicon pools were analyzed using an Agilent Bioanalyzer (Advanced Analytics, Ames, IA, USA) to confirm size selection and quality. Lastly, the amplicon pool was sequenced using an Illumina MiSeq (Illumina Inc., San Diego, CA, USA) with version 3 chemistry.
Sequence data derived from the sequencing process were trimmed using the the open access FASTX-toolkit (v 0.0.14), and demultiplexed and analyzed with the open access bioinformatics software QIIME 1.8.0. Briefly, bacterial and archaeal sequences were clustered into operational taxonomic units (OTUs) using the default UCLUST closed-reference OTU picking algorithm (pick_closed_reference_otus.py) against the curated Greengenes 13_8 reference OTU database (97% similarity threshold), which discards sequences that do not match to the reference database [25,26]. Fungal OTUs were generated using the open reference OTU picking algorithm (pick_open_reference_otus.py) against the UNITE OTUs ITS 12_11 reference database [27]. Singletons (OTUs observed fewer than two times) and OTUs that represented less than 0.005% of the total observations were removed [28] before rarefying the bacterial sequences to an even sampling depth of 8227 sequences per sample to assess alpha (observed OTUs and phylogenetic diversity) and beta diversity (weighted and unweighted Unifrac distances) measures as well as the impact of treatment on OTU abundances that that represented ≥0.5% of the total sequences.

Statistics
Sequence percentages were analyzed using the mixed models procedure of SAS (version 9.4; SAS Institute, Inc., Cary, NC, USA) with treatment as a fixed effect and participant and period as random effects. Post hoc Dunnets adjustments were used to control for multiple comparisons.
To determine if the presence or absence of almonds affected the fecal microbiota, contrast statements were utilized to compare almond (pooled whole, whole roasted, chopped, and butter) to the control diet period. To assess the effect of processing on the microbiota, each almond group (e.g., whole, whole roasted, chopped, or butter) was contrasted against control. The UNIVARIATE procedure and Shapiro-Wilk statistic were used to test for data normality, and log transformations were utilized as needed. The Wilcoxon test was used for fungal count analyses. A probability of p ≤ 0.05 was accepted as statistically significant, and p ≤ 0.10 was considered a trend.

Results
All 18 participants, 10 men and eight women, that were randomized completed the study. Participants had a mean BMI of 29.7 + 4.4 kg/m 2 and an age of 56.7 + 10.2 years (Table 1).
Archaeal sequences accounted for approximated 8.4% of total sequences. Following demultiplexing and quality filtering, the mean and standard deviation of archaeal OTUs identified using closed reference OTU picking was 3081 ± 5289, with a maximum of 26,030 counts per sample and a minimum of zero counts per sample. Treatment did not impact archaea relative abundances. Fungi accounted for approximately 0.1% of total sequences; the mean and standard deviation was 256 + 560 OTUs per sample. The minimum was one OTU/sample (12 individuals), and the maximum was 2410 sequences. Study participants had more (p < 0.05) fungal OTUs when consuming the whole almonds treatment compared to the other treatment groups.
To determine if the presence or absence of almonds in the diet affected the fecal microbiota, contrast statements were utilized to compare almond (pooled whole, whole roasted, chopped, and butter) to the control diet period. The data revealed that almond consumption decreased the relative abundance of the Actinobacteria phylum, as well as Bifidobacterium and Parabacteroides (p < 0.05). The relative abundance of Lachnospira, Roseburia, Clostridium, and Dialister (p ≤ 0.05) increased when the almonds were consumed compared to the control (Table 2). To assess the effect that processing has on the microbiota, each almond group (e.g., whole, whole roasted, chopped, or butter) was contrasted against the control. These results revealed that processing differentially impacted the microbiota. Compared to the control, chopped almonds increased the relative abundance of Lachnospira, Roseburia, and Oscillospira (p < 0.05); whole roasted almonds increased (p = 0.03) the relative abundance of Lachnospira; and whole raw almonds increased (p = 0.007) the relative abundance of Dialister (Table 3). There were no differences between control and almond butter. Almond consumption did not affect bacterial alpha diversity or beta diversity measures (Supplemental Figures S1-S4).    1 Values are least-square means ± pooled SEM; n = 15-18 participants in a crossover design. 2 Values are mean log-normalized sequence abundances ± SE. Treatment effects of genera representing ≥0.5% of the total sequences were evaluated by using a mixed-model ANOVA with post-hoc Dunnet's adjustments for multiple comparisons. Comparisons were made to the control condition. * p ≤ 0.05 for comparison between the respective treatment condition and control condition;ˆp ≤ 0.1 for comparison between the respective treatment condition and control condition.

Discussion
Herein, we report that almond consumption induced changes in the relative abundance of microbes present within the human gastrointestinal tract and that the degree of almond processing differentially impacted the relative abundance of fecal bacterial genera. With regard to specific microbes, the results from this randomized controlled dietary intervention that provided 42 g/d of almonds increased the relative abundance of Clostridium clusters IV and XIVa, including Roseburia, Clostridium, Lachnospira compared to the control diet period that was devoid of almonds. Comparisons between each of the four types of almonds (whole, whole roasted, roasted chopped, and butter) and control revealed that roasted chopped almonds increased the relative abundances of Roseburia, Lachnospira, and Oscillospira. Consumption of whole roasted almonds also increased the relative abundance of Lachnospira. There were no differences in the microbiota between the control group and the almond butter group.
We previously reported that almond processing affected metabolizable energy, with whole natural, whole roasted, and roasted chopped almonds providing up to 25% less metabolizable energy than predicted by calculated Atwater values [19]. Herein, the secondary analyses conducted on the same fecal samples from that previous study provide evidence that food form also impacts the composition of the human intestinal microbiota. Indeed, consumption of roasted, chopped almonds increased the relative abundances of Roseburia, Lachnospira, and Oscillospira compared to control. Whole roasted almonds also increased the relative abundance of Lachnospira, however, they only tended to increase Roseburia. Similarly, there was only a trend for increased relative abundances of Lachnospira and Roseburia with whole, natural almonds. However, whole, natural almonds increased the relative abundance of Dialister. As the whole natural almonds were the hardest and generated a lower number of particles than chopped roasted and roasted whole almonds, this may have not only reduced the metabolizable energy but also the bioaccessibility of nutrients for microbial metabolism. Additional evidence for the relationship between metabolizable energy and the microbiota is provided through the comparisons between almond butter and control. Indeed, disruption of the plant cell wall by roasting and grinding the almonds into almond butter resulted in similar microbial profiles to that of the control period, which is consistent with our previously reported finding that the measured metabolizable energy of almond butter was not different from the calculated value [19]. Overall, these results indicate that roasting, which reduces the fracture force, hardness and subsequently particle size of almonds following mastication [29], as well as the metabolizable energy [19], are also relevant factors that affect the human gastrointestinal microbiota.
Our findings are supported by previous clinical research that reported that up to 82 g/day of almonds did not affect alpha-diversity but did increased the abundance of several Clostridium spp. [30]. Similarly, a clinical trial assessing walnut consumption revealed that 42 g/day of walnuts increased Roseburia, Clostridum and Dialister, and reduced the relative abundances of Bifidobacterium [31]. Although there are not preclinical studies investigating the impact of almond consumption on gastrointestinal microbiota, our results are supported by a rodent study demonstrating that walnuts increased the relative abundances of Firmicutes, including Clostridium [32], Roseburia, and Oscillospira [33]. In vitro work has revealed that finely ground almonds increase butyrate concentrations [20]. Roseburia spp. produce high levels of butyrate in vitro, possess the butyryl coenzyme A (CoA): acetate CoA transferase, and may account for a significant proportion of butyrate-producing bacteria in the human colon [34][35][36]. Thus, the increase in the relative abundance of Roseburia following almond consumption may be important because reduced butyrate-producing bacteria has been linked to metabolic diseases including obesity [37], type 2 diabetes [2], and cardiovascular disease [38]. In addition, as our results revealed an increase in the relative abundance of Roseburia spp., we can hypothesize that these changes in the microbiota may have increased colonic butyrate concentrations in the study participants. However, additional human studies on almond consumption that measure fecal butyrate concentrations are needed to test this hypothesis as well as determine if changes in the gastrointestinal microbiota also contribute to metabolic health outcomes including adiposity, glycemia, and blood cholesterol concentrations.
In addition to providing a good source of dietary fiber, almonds are also rich in unsaturated fatty acids. Unsaturated fatty acids, as opposed to saturated fatty acids, have been shown to have antimicrobial properties through the inhibition of bacterial fatty acid synthesis and growth [39], and also reduce microbial adhesion to mucus [40]. Thus, the increase in the relative abundance of Roseburia spp. following almond consumption in the present study may be related to their ability to metabolize fatty acids. Consumption of polyunsaturated fatty acids may modulate the composition of the gastrointestinal microbiome due to the differing abilities of bacteria to metabolize long-chain fatty acids. For example, in vitro studies have reported that Roseburia spp. are among the most active linoleic acid metabolizing bacteria in the human intestine [12][13][14].
Although the study participants' age range was fairly broad, ranging from 32-78 years of age, 15 of the 18 study participates were 50 years of age or older. Therefore, the gastrointestinal microbiota characteristics of the participants in the current study would generally be reflective of a slightly older population. As the relative abundance of Roseburia spp. has been reported to be negatively associated with age [41,42], almond consumption may help delay age-related microbiota changes. However, as this study was fairly short-term (three weeks per diet period, 20 weeks total), additional longitudinal research on almond consumption and the microbiota in the context of aging is necessary.
The strength of this study is the use of a complete feeding, randomized controlled, cross-over trial with washout periods between each treatment to assess the impact of almond processing on the human gastrointestinal microbiota. Study limitations include lack of bacterial fermentative end-products such as the short-chain fatty acids acetate, propionate, and butyrate. In summary, almond consumption and the form of the almonds differentially impacted the human gastrointestinal microbiota. Nutrients that might not be available to the body through digestion in the upper intestine due to the food matrix can become substrates for microbiota metabolism in the large intestine. In this study of different forms of almonds, when these nutrients enter the large intestine, these substrates alter the composition of the microbiota. These results suggest the gastrointestinal microbiota may contribute to the underlying mechanisms of the beneficial health effects of almond consumption.