**Nutrition, Microbiota and Noncommunicable Diseases**

Editor

**Julio Plaza-D´ıaz**

MDPI • Basel • Beijing • Wuhan • Barcelona • Belgrade • Manchester • Tokyo • Cluj • Tianjin

*Editor* Julio Plaza-D´ıaz Univ Granada Biomed Res Ctr, Inst Nutr & Food Technol Jose Mataix Spain

*Editorial Office* MDPI St. Alban-Anlage 66 4052 Basel, Switzerland

This is a reprint of articles from the Special Issue published online in the open access journal *Nutrients* (ISSN 2072-6643) (available at: https://www.mdpi.com/journal/nutrients/special issues/ Microbiota Noncommunicable Diseases).

For citation purposes, cite each article independently as indicated on the article page online and as indicated below:

LastName, A.A.; LastName, B.B.; LastName, C.C. Article Title. *Journal Name* **Year**, *Article Number*, Page Range.

**ISBN 978-3-03936-916-4 (Hbk) ISBN 978-3-03936-917-1 (PDF)**

c 2020 by the authors. Articles in this book are Open Access and distributed under the Creative Commons Attribution (CC BY) license, which allows users to download, copy and build upon published articles, as long as the author and publisher are properly credited, which ensures maximum dissemination and a wider impact of our publications.

The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons license CC BY-NC-ND.

### **Contents**





### **About the Editor**

**Julio Plaza-D´ıaz** has a degree in pharmacy (2008) and PhD in biochemistry and molecular biology (2014) from the University of Granada, Spain. He has been a researcher at the University of Granada since 2011 and has participated in more than 10 funded research projects. He is the author or coauthor of 50 peer-reviewed publications in the field of identification, characterization, and evaluation of new probiotic/microbiome studies and its impact on health.

### *Editorial* **Nutrition, Microbiota and Noncommunicable Diseases**

#### **Julio Plaza-Diaz 1,2,3**


Received: 21 June 2020; Accepted: 29 June 2020; Published: 2 July 2020

The advent of new sequencing technologies has inspired the foundation of novel research to ascertain the connections between the microbial communities that reside in our gut and some physiological and pathological conditions. The microbiota, defined as the full collection of microbes (bacteria, fungi, and viruses, among others) that naturally exist within a particular biological niche, is estimated to contain 500–1000 species [1–4].

This Special Issue of *Nutrients*, "Nutrition, Microbiota, and Noncommunicable Diseases" contains 13 original publications and seven reviews investigating the contribution of intestinal microbiota on relevant health outcomes in a variety of populations, and animal studies which suggest the growing and extensive interests of research on this topic.

Seven studies were published examining the changes in intestinal microbiota in the human population. Two of these studies recruited patients with metabolic syndrome. Tenorio-Jimenez et al. [5] reported the anthropometric variables and biochemical and inflammatory biomarkers as well as the gastrointestinal microbiome composition changes in a randomized, crossover, placebo-controlled, single-center trial in adult patients newly diagnosed with metabolic syndrome treated either with *Lactobacillus reuteri* V3401 or a placebo during 12 weeks. *L. reuteri* V3401 administration improved selected inflammatory parameters and modified the gastrointestinal microbiome, especially *Verrucomicrobia* [5], and Bellikci-Koyu et al. [6] investigated the effects of regular kefir consumption on gut microbiota composition, and their relation with the components of metabolic syndrome in a parallel-group, randomized, controlled clinical trial for 12 weeks. Gut microbiota analysis showed that regular kefir consumption resulted in a significant increase only in *Actinobacteria* abundance [6].

In two more additional studies, one with healthy elderly women and another with patients with non-alcoholic fatty liver disease (NAFLD), Morita et al. [7] examined the effect of an exercise intervention (12 weeks, trunk muscle training or aerobic exercise training) on the composition of the intestinal microbiota in healthy elderly women. *Bacteroides* abundance was significantly increased only in the aerobic exercise group, particularly in subjects showing increases in the time spent in brisk walking [7], and Chong et al. [8] determined whether inulin supplementation after brief metronidazole therapy is effective in reducing alanine aminotransferase and maintaining weight loss achieved through a very-low-calorie diet among people with NAFLD. Treatment decreased the ratio of Firmicutes/Bacteroidetes [8].

Lau et al. [9] evaluated the association of probiotic ingestion with obesity, type 2 diabetes, hypertension, and dyslipidemia using data from the National Health and Nutrition Examination Survey, 1999–2014. Probiotic supplementation or yogurt consumption were associated with a lower prevalence of obesity and hypertension [9]. In another study with humans, Dalla Via et al. [10] verified whether trimethylamine-N-oxide urinary levels may be associated with the fecal relative abundance of specific bacterial taxa and the bacterial choline trimethylamine-lyase gene *cutC* in human fecal

samples. Correlation analysis showed that the cut-Kp gene cluster was significantly associated with *Enterobacteriaceae* [10].

Finally, in one study with the pediatric population, Kong et al. [11] reported both oral and intestinal microbiota in patients with autism spectrum disorder and controls, with specific microbial patterns [11].

Regarding animal studies, six studies were published examining the changes in intestinal microbiota. Probiotic supplementation, high-fat diet, use of anorexic mice, fiber, and soy intake and antihypertensive effect in metabolomics profiles were analyzed in these studies. Valcarce et al. [12] reported the effect of a short-time probiotic supplementation consisting of a mixture of two probiotic bacteria with proven antioxidant and anti-inflammatory activities on zebrafish sperm quality and male behavior [12]. Hsu et al. [13] examined the alterations of gut microbiota, mediation of short-chain fatty acids (SCFAs) and their receptors, and downregulation of nutrient-sensing signals effects in rats that received a high-fat diet. Increased Firmicutes to Bacteroidetes ratio, *Akkermansia* and *Verrucomicrobia*, and reduced abundance in the genus *Lactobacillus* were associated with blood pressure elevation [13]. Dominique et al. [14] investigated the role of the microbiome and the ClpB protein in the deregulation and self-maintenance of anorexia pathology in mice. Plasma concentration of ClpB was increased in both limited food access and activity-based anorexia mice and it was correlated with the proportion of Enterobacteriaceae in the animal feces [14]. Sasaki et al. [15] investigated the effects of fiber intake timing on metabolism. Data have suggested that inulin is more easily digested by fecal microbiota during the active period than the inactive period. Inulin consumption at breakfast has a greater effect on the microbiota [15]. Tamura et al. [16] investigated soy protein intake effects on intestinal microbiota. Soy protein intake whether in the morning or evening led to a greater microbiota diversity and a decrease in cecal pH resulting from SCFA production compared with casein intake [16]. Finally, Ahn et al. [17] investigated the metabolomics changes in rats that received amlodipine. Serum levels of phosphatidylcholine, lysophosphatidylcholine, sphingomyelin, triglycerides with large numbers of double bonds, cholesterol, sterol derivatives, and cholesterol esters were increased. Amlodipine-induced compositional changes in the gut microbiota are a causal factor in inflammation [17].

Seven reviews investigating the impact of intestinal microbiota on relevant health outcomes in a variety of populations were published. Hills Jr. et al. [18] described a general vision about the gut microbiome and its important role in human health. Salli et al. [19] reported the health benefits of xylitol. The other reviews have described the intestinal microbiota changes in specific conditions, early infancy, hepatic ischemia-reperfusion and regeneration in liver surgery, vaginal microbiota, and cardiovascular diseases. Mesa et al. [20] reported the microbiome changes and how those modulate the inflammatory mechanisms related to physiological and pathological processes that are involved in the perinatal progress. Cornide-Petronio et al. [21] summarized the role of starvation, supplemented nutrition diet, nutritional status, and alterations in microbiota on hepatic ischemia/reperfusion and regeneration. Barrientos-Duran et al. [22] examined the most important aspect in the vaginal microbiota, with special emphasis in bacterial vaginosis, and the maintenance of eubiosis, and Sanchez-Rodriguez et al. [23] discussed how external factors such as dietary and physical activity habits influence host microbiota and atherogenesis, the potential mechanisms of the influence of gut microbiota in host blood pressure, and the alterations in the prevalence of those bacterial genera affecting vascular tone and the development of hypertension. Finally, Plaza-Diaz et al. [24] revisited the effects of sweeteners on gut microbiota.

The present Special Issue provides a summary of the progress on the topic of intestinal microbiota and its important role in human health in different populations, which will be of interest from a clinical and public health perspective. Nevertheless, more studies with more samples and comparable methods are necessary to understand the actual function of intestinal microbiota in disease development and health maintenance.

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

**Acknowledgments:** Julio Plaza-Diaz is part of the University of Granada, Plan Propio de Investigación 2016, Excellence actions: Units of Excellence; Unit of Excellence on Exercise and Health (UCEES).

**Conflicts of Interest:** The author declares no conflict of interest.

#### **References**


© 2020 by the author. 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*

### **Randomised Double-Blind Placebo-Controlled Trial of Inulin with Metronidazole in Non-Alcoholic Fatty Liver Disease (NAFLD)**

**Clara Yieh Lin Chong 1, David Orr 2,\*, Lindsay D. Plank 3, Tommi Vatanen 1, Justin M. O'Sullivan <sup>1</sup> and Rinki Murphy 4,\***


Received: 5 March 2020; Accepted: 24 March 2020; Published: 27 March 2020

**Abstract:** *Background*: Non-alcoholic fatty liver disease (NAFLD) can be ameliorated by weight loss although difficult to maintain. Emerging evidence indicates that prebiotics and antibiotics improve NAFLD. *Aim*: To determine whether inulin supplementation after brief metronidazole therapy is effective in reducing alanine aminotransferase (ALT) and maintaining weight loss achieved through a very-low-calorie diet (VLCD) among people with NAFLD. *Methods*: Sixty-two people with NAFLD commenced 4-week VLCD using Optifast meal replacements (600 kcal/day). Sixty were then randomised into a 12-week double-blind, placebo-controlled, parallel three-arm trial: (1) 400 mg metronidazole twice daily in Week 1 then inulin 4 g twice daily OR (2) placebo twice daily in week one then inulin OR (3) placebo-placebo. Main outcomes were ALT and body weight at 12 weeks. Fecal microbiota changes were also evaluated. *Results*: Mean body mass index (BMI) and ALT reduced after VLCD by 2.4 kg/m2 and 11 U/L, respectively. ALT further decreased after metronidazole-inulin compared to after placebo-placebo (mean ALT change −19.6 vs. −0.2 U/L, respectively; *p* = 0.026); however, weight loss maintenance did not differ. VLCD treatment decreased the ratio of *Firmicutes*/*Bacteroidetes* (*p* = 0.002). *Conclusion*: Brief metronidazole followed by inulin supplementation can reduce ALT beyond that achieved after VLCD in patients with NAFLD.

**Keywords:** prebiotics; alanine aminotransferase; antibiotic; Optifast; gutmicrobiome; inulin;metronidazole

#### **1. Introduction**

Non-alcoholic fatty liver disease (NAFLD) is defined by the pathological accumulation of fat in the liver and is now the leading cause of chronic liver disease [1]. NAFLD encompasses a spectrum of diseases ranging from simple fatty liver (steatosis) through to non-alcoholic steatohepatitis (NASH), which, in turn, leads to fibrosis, irreversible cirrhosis and, finally, hepatocellular carcinoma (HCC) in a small proportion of people [2,3]. The milder simple steatosis is characterised by the ectopic accumulation of fat in the liver, usually associated with energy-surplus-induced obesity. It is believed that multiple parallel factors (diet, insulin resistance, mitochondrial dysfunction and inflammation), acting synergistically in genetically predisposed individuals, are implicated in the development and progression of NAFLD.

An accumulating number of animal and human studies suggest a compelling role for gut microbiota in NAFLD, which is both transmitted by gut microbiota and reversed by a combination of ciprofloxacin and metronidazole antibiotics in animal models [4]. NAFLD is associated with dysbiosis of the gut microbiota, which is thought to lead to increased gut permeability, and abnormal choline and bile acid metabolism, leading to inflammation and increased hepatic fat accumulation [5]. An indication of the involvement of gut microbiota in NAFLD development was first apparent when hepatic steatosis developed in patients undergoing jejunal–ileal bypass surgery, coinciding with intestinal bacterial overgrowth in the blind loop. The hepatic steatosis regressed once patients were treated with the antibiotic metronidazole [6] which is commonly used for the treatment of small intestinal bacterial overgrowth [7]. While small intestinal bacterial overgrowth has been shown to be more prevalent in NAFLD [8–11], antibiotic treatment of NAFLD has not been investigated due to concerns about long-term use being associated with side effects, antimicrobial resistance and uncertain efficacy.

The cornerstone of NAFLD treatment currently is to offer lifestyle advice that targets 7% to 10% weight loss and is proven to be effective [12,13]. Recent evidence shows that very-low-calorie diets (VLCDs) [14] and bariatric surgery [15] are very effective in achieving weight loss and remission of associated comorbidities. Both these strategies alter gut microbiota, but to a lesser extent after dietary modification than after surgery [16–18]. However, the maintenance of weight loss remains a challenge and better alternatives to targeting specific mechanistic dysfunction are needed.

Prebiotics, which are nondigestible food ingredients that are fermented in the gut and modulate microbiota in a favourable way for the host, have shown promise in the treatment of NAFLD. A systematic review of 26 randomised controlled trials investigating the metabolic benefits of prebiotics concluded that prebiotics improve satiety, postprandial glucose and insulin in both healthy and obese individuals [19]. A meta-analysis of nine randomised controlled trials in NAFLD showed a reduction in body mass index (BMI) and an overall improvement in aminotransferase (ALT) with the use of prebiotics [20]. However, the use of a combination of strategies targeting gut microbiota dysbiosis of NAFLD such as VLCD, metronidazole and prebiotics in succession has not previously been investigated.

We hypothesised that the beneficial metabolic effects of short-term VLCD among adults with NAFLD could be enhanced by the brief use of metronidazole to target dysbiotic gut microbiota followed by a period of inulin supplementation to maintain this. We conducted a single centre, randomised, placebo-controlled, double-blind clinical three-arm trial of 12 weeks of inulin supplementation with or without an initial week of metronidazole cotreatment among adults with NAFLD who had all received four weeks of VLCDs.

#### **2. Patients and Methods**

#### *2.1. Study Design*

This study focused on adults with an established diagnosis of NAFLD attending Auckland City Hospital hepatology outpatient clinic. Patients either had histological evidence of NAFLD based on a liver biopsy, a phenotypic diagnosis based on the presence of BMI > 27 kg/m<sup>2</sup> and type 2 diabetes or metabolic syndrome (WHO criteria) with an elevated ALT (male > 40 U/L, female > 30 U/L) and age >18 years and <75 years. Exclusion criteria were alcohol consumption of more than 20 g per day for at least 3 consecutive months during the previous 5 years as assessed by a questionnaire. Participants were excluded if they had cirrhosis, hepatitis C or another liver disease, if they were awaiting or had previous bariatric surgery, had an allergy to eggs, nuts or metronidazole, a history of drug and alcohol abuse, a calculated eGFR less than 60 mL/min (MDRD formula) or current participation in other therapeutic trials. Ethics approval was from Health and Disability Ethics Committee NTX/12/05/040/AM02; ANZCTR registration number: 12613001002774, prospectively registered on 10 September 2013.

#### *2.2. Randomisation and Treatment Groups*

Sixty-two participants with NAFLD who met all eligibility criteria and provided written informed consent were provided with 3 Optifast meal replacements (600 kcal/day) per day for 4 weeks to initiate weight loss after which the 60 participants who attended the second study visit were randomly assigned to one of three parallel groups (1:1:1; Figure A1). The metronidazole and inulin group (Group MI) received metronidazole (dose of 400 mg twice daily for 7 days) along with inulin (at a dose of 4 g twice daily for 12 weeks); the placebo and inulin group (Group PI) received metronidazole-like placebo (twice daily for 7 days) along with inulin (at a dose of 4 g twice daily for 12 weeks); the placebo and inulin placebo group (Group PP) received metronidazole-like placebo (twice daily for 7 days) along with inulin-like placebo (containing maltodextrin at a dose of 4 g twice daily for 12 weeks).

The inulin dose was selected on the basis of previous prebiotic studies and was provided by Cargill Belgium. A metronidazole dose of 400 mg twice daily was selected as slightly lower than the standard dose of 400 mg three times daily used for various medical conditions, such as bacterial vaginosis, dental abscess and giardiasis, for increased adherence than three times daily. Metronidazole and matching placebo-containing maltodextrin were encapsulated by the Auckland Hospital Clinical Trials' Pharmacy department. All participants, their health care providers and assessment staff were blinded to treatment allocation. Participants were asked to take the inulin/matching placebo powder twice daily before breakfast and before dinner using a 4 g measuring spoon and two level spoonfuls dissolved into water. All participants were given a standardised set of recommendations about lifestyle changes and diet following the initial expected weight loss period during VLCD at time of study randomisation

In total, there were four time points in this study: baseline (study entry), Week 4 (after 4 weeks VLCD pre-randomisation), Week 16 (post-randomisation, at the end of treatment) and Week 28 (post-treatment follow up phase to evaluate whether there were any persistent effects detected beyond the treatment period) as shown in Figure 1. All participants underwent assessment for body weight, height, waist and hip circumference at each of these 4 timepoints. Blood samples for assessment of fasting lipids, glucose, insulin and liver Fibroscan CAP were obtained at baseline, Week 4 and then Week 16.

**Figure 1.** Assessment and sample collection timeline.

#### *2.3. Stool Sample Collection*

Stool samples were collected at each time point (Figure 1): baseline, Week 4, Week 16 and Week 28. Study participants collected the stool samples at home, using a sterile collection tube, prior to their hospital visits. Stool samples were stored at −70 ◦C from the beginning of the study (2013/2015) until DNA extraction was performed (2017).

DNA was extracted from stool samples using the QIAamp® Fast DNA Stool Mini Kit according to the manufacturer's protocol. Extracted DNA quality and quantity were measured using a NanoPhotometer N60 (IMPLEN, Germany; Table S1) and a Qubit (Invitrogen, US).

#### *2.4. 16S rRNA Gene Amplicon Sequencing*

Extracted DNA (mean yield = 6733.4 ng; mean 260/280 = 1.97; mean concentration = 33.7 ng/μL) was sent to the School of Biological Sciences (The University of Auckland, New Zealand) for 16S rRNA amplicon sequencing on an Illumina MiSeq sequencing platform. Sequences are available from SRA project number SUB5068044. Then, 16S rRNA gene amplicon sequencing (16S sequencing) libraries were prepared using the Nextera XT kit (Illumina). V3 and V4 regions were targeted for 16S sequencing by using the 16S Amplicon PCR Forward Primer (TCGT CGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG) and Reverse Primer (GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC).

All amplicons were sequenced on the Illumina MiSeq 600 cycle run to generate an average of 121,346 sequence reads with paired-end (300 bp each) reads per sample.

#### *2.5. 16S rRNA Amplicon Sequence Analyses*

The 16S sequencing data were processed using QIIME 2 (v. 2018.4) [21]. Briefly, sequence quality control and denoising were performed using DADA2 [22]. The quality control step also included the filtering of PhiX reads and chimeric sequences. The sequences obtained after denoising were then classified using Greengenes 13\_8 release data to identify amplicon sequencing variants (ASVs) for sequences with >99% sequence similarity. Samples that were included in downstream analyses had filtered sequence counts ranging from 12,123 to 109,977 (median 53,071). Three samples with less than 10,000 sequencing reads were removed.

#### *2.6. Primary and Secondary Outcomes*

The primary outcome was the proportion maintaining a ≥7% weight loss at the end of the 12-week variable treatment period compared to their baseline (before the fixed 4-week VLCD treatment period). Secondary outcomes measured at Week 16 (the end of the 12-week variable treatment period) included changes in ALT, glycaemia, lipids, Fibroscan® CAP from what was achieved at Week 4 (after VLCD treatment period) and the changes in gut microbial community from baseline to 28 weeks.

#### *2.7. Statistical Analysis*

The planned sample size for this pilot study was 60 subjects with an equal assignment to each of the three study groups (20 per group). We estimated that, with this sample size, the study would have 80% power to detect a difference in the proportion achieving a sustained weight loss of ≥7% at the end of the 12-week treatment period which we anticipated would be achieved by 50% of those receiving metronidazole and inulin supplemented diet, compared to 5% in the other two placebo-containing groups, with a two-sided type 1 error of 0.05. The primary outcome was assessed using Fisher's exact test. Pre-planned analyses for secondary outcomes were comparisons of the changes over the 12-week variable treatment period in the MI and PI groups with those in the PP group. Two-sample *t*-tests were used for these comparisons for normally distributed data and Mann–Whitney *U*-tests for non-normal data. Within-cohort changes over the VLCD period were analysed using paired *t*-test or Wilcoxon signed-rank test as appropriate. Data are presented as mean (SD) or median (Quartile 1, Quartile 3) for normally and non-normally distributed data, respectively.

#### *2.8. Statistical Methods for Microbiota Analysis*

Omnibus associations between microbial community structure and patient metadata were assessed using Permutational Analysis of Variance (PERMANOVA) (adonis function from the vegan package in R, 10,000 permutations) and Bray–Curtis dissimilarities. Wilcoxon matched-pairs signed-rank tests were used to assess relative abundance across timepoints. Associations between individual microbial taxa and patients' metadata were assessed using Multivariate Association with Linear Models (MaAsLin) [23], controlling for age as a possible confounding factor and repeated sampling per individual by a random effect. The Kruskal–Wallis test was used when more than two independent groups were compared. ASVs that were present in less than 20% of samples were filtered out. *p*-values were corrected for multiple testing using the Benjamini–Hochberg procedure [24,25] and FDR corrected *p*-values (*q*-values) were reported.

#### **3. Results**

#### *3.1. Participants*

Enrollment into this trial occurred between March 2013 and March 2015. Sixty-two participants entered the study and began VLCD, of whom 60 attended Visit 2 and were then randomised: 20 attended metronidazole and inulin (MI), 20 attended metronidazole placebo and inulin (PI) and 20 attended metronidazole placebo and inulin placebo (PP). Participant flow through the trial is shown in Figure A1. The mean age was 50 years (range 19–71), BMI 31.6 kg/m2 (range 25.2–41.9) and ALT 66 U/L (range 30–141). The three groups were well matched with respect to demographic characteristics, clinical and laboratory data at study entry and after four weeks of VLCD (Table A1). Over the VLCD period, there were significant reductions in body weight, waist:hip ratio, blood pressure, ALT and gamma-glutamyl transferase (GGT), total and LDL cholesterol, triglycerides, fasting glucose, HbA1c, CRP and Fibroscan CAP score.

#### *3.2. Primary and Secondary Outcomes*

Of the 62 participants who were assessed at baseline, 60 were randomised and 56/60 (93.3%) participants completed the study. The clinical endpoint of achieving sustained weight loss of ≥7% at 16 weeks compared to baseline pre-VLCD was reached by 55% in group MI compared with 53% in group PI and 35% in group PP. These were not statistically significantly different (*p* = 0.473). At 28 weeks, a sustained weight loss of ≥7% was reached by 42% in group MI, 35% in group PI and 25% in group PP (*p* = 0.584).

Although there was no difference in BMI between the three treatment groups at 12 weeks, only the group receiving inulin with an initial one week of metronidazole, group MI, had a significant, further improvement in ALT (Table A2). However, this group had no significant change in Fibroscan® CAP score or in other markers of metabolic syndrome such as blood pressure, fasting lipids and glycaemia. No cases of adverse events requiring discontinuation of inulin were reported.

#### *3.3. Gut Microbial Changes in Our Study Cohort*

A total of 127 stool samples were obtained for analysis (Figure A2). Patient compliance in providing stool samples was highest at Week 4 with 38 (29.9%) stool samples and lowest at Week 28 with 26 (20.5%) stool samples (Figure A2). All four stool samples were obtained from 10 study participants.

After Optifast VLCD: VLCD treatment explained 5.3% (PERMANOVA, *p* = 0.0024) of the variance in microbial profiles. *Bacteroidetes* and *Firmicutes* were the two most highly represented bacterial phyla in our cohort (Figure 2). After four weeks of VLCD, the relative abundance of *Bacteroidetes* increased (Wilcoxon signed-rank test, *p* = 0.047) while *Firmicutes* decreased (Wilcoxon signed-rank test, *p* = 0.01) (Figure 2). Furthermore, the ratio of *Firmicutes*/*Bacteroidetes* decreased significantly after VLCD treatment (Wilcoxon signed-rank test, *p* = 0.002, *n* = 30) (Figure 3).

**Figure 2.** *Firmicutes* and *Bacteroidetes* are the dominant phyla in the subjects before and after very-low-calorie diet (VLCD) treatment. The figure shows boxplots of five typical human microbiota phyla. The boxes indicate the interquartile range (IQR) while the notch region shows the 95% confidence interval for the median and the whiskers extending from the boxes represent the distribution within 1.5 × IQR, with points beyond this range shown as outliers.

**Figure 3.** The ratio of *Firmicutes*/*Bacteroidetes* phyla decreased (Wilcoxon signed-rank test, *p* = 0.002, *n* = 30) from baseline to Week 4 after the VLCD diet. Boxplots as in Figure 2.

Linear modelling identified three statistically significant genera (*q* < 0.1, Table S2), all belonging to the phylum Firmicutes. *Roseburia*, *Streptococcus* and *Dialister* genera displayed an association with the VLCD treatment and were significantly lower after VLCD treatment compared to the other time points (Figure 4). However, the microbial alpha diversity metrics showed no significant change from baseline following a VLCD diet (Shannon, *p* = 0.968; Wilcoxon signed-rank test, *n* = 30).

**Figure 4.** The relative abundance of genera *Roseburia* (**A**), *Streptococcus* (**B**) and *Dialister* (**C**) were lower (linear mixed-effects model, *q*-value = 0.0005, 0.0005 and 0.03, respectively) after VLCD treatment compared to baseline. Baseline, *n* = 35; Week 4, *n* = 38. Boxplots as in Figure 2.

#### *3.4. Enrichment of Distinct Gut Microbial Profile in Our Study Cohort*

There were no significant differences in alpha diversity (Shannon's diversity index) at Week 16 between intervention groups (*p* = 0.755, Kruskal–Wallis test). Similarly, comparison between Week 4 and Week 16 found no significant difference between groups (*p* = 0.949, Kruskal–Wallis test).

Linear modelling identified three taxa, genera *Roseburia*, *Anaerotruncus* and family *Lachnospiraceae*, all belonging to the phylum Firmicute, were associated to the antibiotic/prebiotic treatment period (*q* = 0.026) (Table S3). However, comparison between groups found no differences (*p* = 0.097, Kruskal–Wallis test).

Linear modelling revealed a suggestive association between genus *Turicibacter* and plasma ALT levels (*q*-value = 0.086, Table S4) when corrected for patient age. However, neither linear modelling nor PERMANOVA identified any significant associations between microbial taxa and plasma ALT in the cross-sectional model at Week 16 (*n* = 28 samples).

#### **4. Discussion**

There are currently few approved treatment options for NAFLD patients beyond dietary measures to lose weight. In this study, four weeks of VLCD resulted in a significant weight loss. Subsequent inulin supplementation for 12 weeks, with or without an initial one week of metronidazole), did not improve weight loss maintenance. The transition to a real food diet after a period of VLCD meal replacements is usually associated with weight regain and recurrence of NAFLD. Indeed, only 48% of participants were able to maintain a ≥7% weight loss after 12 weeks of transition to a real food diet. Despite similar weight loss maintenance, the group who received metronidazole and inulin (after the initial VLCD period) achieved a further significant reduction in ALT. The reduction in ALT suggests reduced steatohepatitis, which, surprisingly, in our study—in contrast to other prebiotic studies that have shown an improvement in ALT commensurate with weight loss [20]—occurred without further weight loss. This finding supports the potential role of metronidazole in improving steatohepatitis through the treatment of intestinal bacterial overgrowth and/or through altering gut microbial functions that enhance efflux of free fatty acids and de novo lipogenesis in the liver. These mechanisms can occur without weight loss as the cause and are not necessarily apparent by simple characterisation of microbial abundances in the faeces. This is because there is substantial inter-individual variability in gut microbiota among patients with NAFLD and bacterial abundance in faecal samples do not directly demonstrate activity or metabolite production of the taxa present in the small intestine.

Nonetheless, four weeks of VLCD (Optifast) had a major effect on decreasing the ratio of *Firmicutes*/*Bacteroidetes* in faeces, as well as decreasing the abundance of genera *Roseburia*, *Streptococcus* and *Dialister*. The genus *Roseburia*, a member of clostridial cluster XIVa [26], consists of obligate Gram-positive anaerobic bacteria and is an important butyrate-producing colonic bacterium [27–30] and suggested to be able to alleviate inflammation by stimulating Treg cell differentiation [31,32]. Butyrate is a short-chain fatty acid produced mainly by the enteric microbiome [33,34]. It is a crucial element in the normal development of colonic epithelial cells [35] and preferred energy source in the colonic mucosa [36]. A previous study has shown that a butyrate-producing probiotic MIYAIRI 588 strain of *Clostridium butyricum* effectively improved hepatic indexes in an animal model [33]. Butyrate has also been suggested to confer various beneficial metabolic effects such as enhancing mitochondrial activity [34], increasing insulin sensitivity [37], conveying anti-inflammatory potential [38] as well as increasing intestinal barrier function [39]. However, the role of butyrate in NAFLD is controversial as patients with NASH were shown to have higher faecal butyrate compared to healthy subjects [40]. *Roseburia* has also been detected as significantly elevated in NAFLD patients compared to healthy controls [41]. This genus is suggested to be one of the gut microbiota biomarkers that is shared by obese patients with metabolic disease and is negatively associated with body mass index (BMI) [32]. Ironically, a depletion of *Roseburia* was observed after four weeks of Optifast VLCD treatment, with a significant reduction in BMI in our study. In fact, our observation is similar to Duncan et al. [42] and, subsequently, Alemán et al. [43] who observed a significant reduction of genus *Roseburia* after VLCD intervention. Since *Roseburia* are predominantly polysaccharide-degrading bacteria [44], we postulate that the observed reduction in the genus *Roseburia* is actually due to reduced dietary carbohydrates from VLCD and not directly linked to BMI.

Similarly, the genus *Streptococcus*, a possible biomarker of NAFLD [45] reduced significantly after VLCD treatment compared to baseline (Figure 4). This corroborates previous work where *Streptococcus* was enriched in NAFLD and NAFLD-cirrhosis patients [46,47] compared to both healthy subjects [48,49] and obese individuals [45]. Future studies could investigate metabolic activities and molecular mechanisms linking *Streptococcus* and NAFLD aetiology.

Optifast-based VLCD reduced the alpha diversity (Shannon's diversity index) of gut microbiota seen between baseline to Week 4 (Figure S1). Altered diversity has also been shown to occur in a similar Optifast-based VLCD study of three months duration in 18 obese participants, although changes regressed during the subsequent weight maintenance phase and return to a real food diet [18]. It is noticeable that upon transition to a food diet, metronidazole-inulin and placebo-inulin groups

both shared a similar fluctuation pattern in Shannon's diversity index for all time points compared to the placebo-placebo group which demonstrated a relatively stable pattern.

Due to limited stool sample collection after VLCD, lack of statistical power precluded the assessment of any gut microbial differences in abundance between metronidazole treated and non-metronidazole treated groups. Further, we cannot rule out the possibility that low compliance in the stool sample collection may have added further bias in microbiome analysis. However, linear modelling revealed that the genus *Turicibacter* was associated with the plasma ALT levels within placebo-inulin and placebo-placebo groups. *Turicibacter* has been suggested to be responsive to the cholesterol level in the diet [50]. This is in line with the fact that hepatic free cholesterol accumulation and altered cholesterol homeostasis will lead to liver injury and eventually contribute to the pathogenesis of NAFLD/NASH [51]. We suggest an association between *Turicibacter* and plasma ALT levels which clearly need further research.

Finally, maltodextrin may not have been as inert as a placebo should be, given there is some evidence that maltodextrin detrimentally impacts the intestinal environment [52–54]. However, most of these studies used much higher doses of maltodextrin than was used in the form of placebo to match inulin in this study.

#### **5. Conclusions**

In conclusion, this is the first clinical trial evidence that supplementation with prebiotic inulin following brief metronidazole therapy can further reduce ALT after four weeks of VLCD therapy in patients with NAFLD. A prominent shift in phyla Firmicutes, Bacteroidetes, genera *Roseburia*, *Streptococcus* and *Dialister* were seen after four weeks of Optifast treatment. Unfortunately, limited stool samples collection after VLCD treatment resulted in insufficient power to detect a significant difference in gut microbiota with additional metronidazole or inulin. Nevertheless, a potential role of metronidazole, together with inulin in altering the gut microbial function (e.g., metabolites production), is suggested to alleviate steatohepatitis, as evidenced by the reduction of ALT in the MI group. Future studies are recommended to examine the effect on microbial metabolites which does not manifest in measuring the diversity of microbiota. Furthermore, this clinical therapeutic approach requires validation in larger clinical studies with the possibility of low compliance on stool samples collection is accommodated.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6643/12/4/937/s1, Figure S1: Violin plot on the Shannon's diversity index for each group on each time point, Table S1: DNA quality and quantity measured using a NanoPhotometer N60, Table S2: Linear modelling results showing genera associated with the VLCD treatment, Table S3: Linear modelling results showing genera associated with the antibiotic/prebiotic treatment period, Table S4: Linear modelling results showing suggestive association between genus *Turicibacter* and plasma ALT levels.

**Author Contributions:** Conceptualisation, D.O. and R.M.; data curation, C.Y.L.C. and T.V.; formal analysis, C.Y.L.C., D.O., L.D.P., T.V. and J.M.O.; funding acquisition, D.O. and R.M.; investigation, C.Y.L.C., D.O., L.D.P., J.M.O. and R.M.; methodology, D.O., L.D.P., T.V., J.M.O. and R.M.; project administration, D.O.; resources, J.M.O. and R.M.; Supervision, D.O., T.V., J.M.O. and R.M.; visualisation, C.Y.L.C., T.V. and J.M.O.; writing—original draft, C.Y.L.C. and R.M.; writing—review and editing, C.Y.L.C., D.O., L.D.P., T.V., J.M.O. and R.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by A+ Trust, New Zealand Society of Gastroenterology.

**Acknowledgments:** The inulin used in this work was provided gratis by Cargill Belgium. We are grateful to Gil Hardy for overseeing the packaging of inulin and maltodextrin performed by AnQual Laboratories. We thank Auckland Pharmacy Clinical Trials for the encapsulation of the metronidazole and placebo tablets.

**Conflicts of Interest:** The authors declare no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **Abbreviations**


#### **Appendix A**

**Table A1.** Baseline and week 4 characteristics of study participants.



**Table A1.** *Cont.*

Data are mean (SD), median (Q1, Q3) or number of patients. \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001 (paired *t*-test or Wilcoxon signed rank test).

#### **Appendix B**

**Table A2.** Changes in characteristics at week 16, after 12 weeks of treatment following randomisation to metronidazole/inulin, placebo/inulin or placebo/placebo.


Data are mean (SD) or median (Q1, Q3).

#### **Appendix C**

**Figure A1.** Flow diagram showing different phases of the study and number of participants at each stage. EOT: end of treatment; MI: metronidazole-inulin; PI: placebo-inulin; PP: placebo-placebo.

#### **Appendix D**

**Figure A2.** Flow diagram showing total stool sample collection at different time points from different groups. MI: metronidazole-inulin; PI: placebo-inulin; PP: placebo-placebo.

#### **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* **Relationship Between the Gastrointestinal Side E**ff**ects of an Anti-Hypertensive Medication and Changes in the Serum Lipid Metabolome**

#### **Yoomin Ahn 1, Myung Hee Nam <sup>2</sup> and Eungbin Kim 1,\***


Received: 28 November 2019; Accepted: 6 January 2020; Published: 13 January 2020

**Abstract:** An earlier study using a rat model system indicated that the active ingredients contained in the anti-hypertensive medication amlodipine (AMD) appeared to induce various bowel problems, including constipation and inflammation. A probiotic blend was found to alleviate intestinal complications caused by the medicine. To gain more extensive insight into the beneficial effects of the probiotic blend, we investigated the changes in metabolite levels using a non-targeted metabolic approach with ultra-performance liquid chromatography-quadrupole/time-of-fligh (UPLC-q/TOF) mass spectrometry. Analysis of lipid metabolites revealed that rats that received AMD had a different metabolome profile compared with control rats and rats that received AMD plus the probiotic blend. In the AMD-administered group, serum levels of phosphatidylcholines, lysophosphatidylcholines, sphingomyelins, triglycerides with large numbers of double bonds, cholesterols, sterol derivatives, and cholesterol esters (all *p* < 0.05) were increased compared with those of the control group and the group that received AMD plus the probiotic blend. The AMD-administered group also exhibited significantly decreased levels of triglycerides with small numbers of double bonds (all *p* < 0.05). These results support our hypothesis that AMD-induced compositional changes in the gut microbiota are a causal factor in inflammation.

**Keywords:** lipid metabolome; amlodipine; probiotics; corticosterone; ACTH; gut bacteriome

#### **1. Introduction**

Many metabolic diseases (e.g., obesity, hypertension, and diabetes) have emerged as serious health problems in developed countries, mainly as a result of changes in eating habits and developments in the food industry. For example, according to a report published in 2017 by the Organization for Economic Co-operation and Development (OECD), the obesity rate in the U.S. was 30.9% in 2000 and increased to 38.2% in 2014. In the case of hypertension, although prevalence decreased from 1999 to 2016, the absolute burden caused by hypertension has increased [1]. Unlike infectious diseases, metabolic disorders are typically chronic and manageable rather than remediable, forcing patients to take medications almost ad infinitum. Along with their beneficial effects, long-term medications may cause some unwanted effects. There is increasing evidence that some side effects of long-term medications, including gastrointestinal (GI) disorders (such as constipation, diarrhea, and irritable bowel syndrome (IBS)), are related to disruption of the gut microbial population, referred to as dysbiosis [2–5]. Many medications cause gut microbiota dysbiosis even though they are not considered antibiotics [6,7].

The human gut microbiota is a complex ecosystem, consisting of approximately 1–4 <sup>×</sup> 1015 microbial cells. The gut microbiota establishes a close relationship with the host through interactions among themselves and with host cells in the GI tract [5,8–10]. Hence, it seems logical rather than surprising that maintenance of a well-balanced gut microbial community is a prerequisite for healthy functioning of the whole system. Indeed, the gut microbiota is proposed to be an essential "organ" that functions to maintain nutrient metabolism, immune function, and metabolic homeostasis [11–14]. Recent studies show that the gut microbiota affects neurodevelopment and diverse brain functions by regulating the gut–brain axis, the bidirectional communication between the brain and the gut [15–18]. Many of these studies reported only correlative or associative findings; however, efforts have been undertaken to examine causality and mechanism in the microbiome.

We previously reported that amlodipine (AMD), the active ingredient in a hypertension medicine, is an aggravating factor in various bowel problems, including constipation and inflammation. This is because it induces compositional changes in the gut microbiota, since normalization of the gut microbiota alleviates intestinal complications caused by AMD [19]. To investigate the effects of the gut microbiome on the host, we performed a comparative analysis of lipid metabolome in serum samples from rats that received saline (null control), AMD, or AMD plus a probiotic blend (AMD+PB). We chose to examine lipid metabolites because they are strongly associated with high blood pressure [20], and AMD is used as a treatment for hypertension.

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

#### *2.1. Experimental Rats*

A total of 18 six-week-old male Sprague Dawley rats were randomly divided into three groups (*n* = 6/group) to receive saline (null control), AMD, or AMD+PB. The probiotic blend (PB) was obtained in powder form and consisted of *Bifidobacterium lactis* CBT BL3 (KCTC 11904BP), *Bifidobacterium longum* CBT BG7 (KCTC 12200BP), *Bifidobacterium bifidum* CBT BF3 (KCTC 12199BP), *Lactobacillus acidophilus* CBT LA1 (KCTC 11906BP), *Lactobacillus rhamnosus* CBT LR5 (KCTC 12202BP), and *Streptococcus thermophilus* CBT ST3 (KCTC 11870BP) (Cell Biotech Co., Ltd., Seoul, Korea). The PB also contained the excipients fructooligosaccharide, lactose, galactooligosaccharide, orange flavor powder, milk flavor powder, Mg-stearate, L-ascorbic acid, vitamin E, dry-formed vitamin A, vitamin B6 hydrochloride, and vitamin B1 hydrochloride. There were approximately equal numbers (ca. 1.67 <sup>×</sup> 109 CFUs/g) of viable cells of each of the six bacterial strains in the PB. The total number of viable cells in the powdered form of the product was determined by measurement to be 1 <sup>×</sup> 1010 CFUs/g, which was diluted in water for oral administration of 1 <sup>×</sup> 107 CFUs/day.

Three rats were housed in a single cage, so two cages were used for each treatment group. After a one-week acclimation period, oral gavage of PB was administered each day in a dose of ~1 <sup>×</sup> 10<sup>7</sup> CFUs. Starting in the third week, AMD was administered to the rats daily for 2 weeks by oral gavage (2 mg/kg/day). The daily dose of AMD was determined as previously described [21]. All rats were housed under the following conditions: temperature 23 ± 1 ◦C, relative humidity 55–65%, and a 12 h light cycle. Metabolic data (weight, food intake, and water intake) were collected every day. Weight data were measured individually for each animal, but food and water intake were measured for each cage rather than for each animal. The use and care of the animals were reviewed and approved by the Institutional Animals Care and Use Committee at the Cell Biotech R&D Centre (CBTJ-15-02). All animal procedures were in accordance with the Guide for the Care and Use of Laboratory Animals issued by the Laboratory Animal Resources Commission of Cell Biotech R&D Centre.

#### *2.2. Serum Collection and Serum Lipid Metabolite Analysis*

On day 28, the rats were sacrificed by CO2 asphyxiation. It should be noted that in the AMD group one rat died before scarification. Blood samples were collected from the heart in micro tubes, kept at 4 ◦C for 1 h, and then centrifuged at 2200× *g* for 15 min. The supernatant was stored at −80 ◦C until use. Each serum sample was prepared by adding 180 μL of isopropyl alcohol to 45 μL of serum (serum:IPA, 1:4) and then vortexing for 1 min. The mixture was incubated at −20 ◦C for 3 h. Then, the samples were centrifuged at 14,000× rpm at 4 ◦C for 20 min. The supernatant was then diluted with an equal

volume of deionized water and injected into an ultra-performance liquid chromatography/quadrupole time-of-flight mass spectrometry (UPLCQ/TOF–MS) machine (Waters Corporation, Milford, MA, USA). The lipid metabolites in the serum were separated using an Acquity UPLC CSH C18 column (2.1 × 100 mm, 1.7 μL particle size; Waters Corporation). The column temperature was 55 ◦C. The mobile phase consisted of acetonitrile:water (60:40) with 10 mM ammonium formate in 0.1% formic acid (A) and isopropanol:acetonitrile (90:10) with 10 mM ammonium formate in 0.1% formic acid (B). The flow rate was set at 0.4 mL/min. The samples were eluted using the following conditions: initial 40% B to 53% at 2 min, to 50% A at 2.1 min, to 54% B at 12 min, to 70% B at 12.1 min, to 1% B at 18 min, to 40% B at 18.1 min, followed by equilibration for an additional 2 min. Mass acquisition was performed in positive and negative electrospray ionization modes. Mass data were collected in the range of *m*/*z* 60–1400 for 20 min with a scan time of 0.25 s and an inter-scan time of 0.02 s. The source and desolvation temperatures were 120 and 550 ◦C, respectively.

#### *2.3. Processing and Analysis of Mass Spectrometry Data*

The Progenesis QI software (Waters Corporation) was used for data processing, including mass ion alignment, normalization, and peak picking. The intensities of the mass peaks for each sample were normalized according to the total ion intensity and Pareto scaled using SIMCA-P+ 12 software (Umetrics, San Jose, CA, USA).

To differentiate among the intensities of the mass peaks in each treatment group, principal component analysis (PCA) was performed. In addition, orthogonal partial least-square discriminant analysis (OPLS-DA) was used for the selection of metabolites.

Metabolites were identified by matching the measured mass spectra with references in the Human Metabolomics Database (http://www.hmdb.ca/) and METLIN (http://metlin.scripps.edu/). Lipids identified in the samples were validated on the basis of isotope similarity and fragmentation patterns. Hierarchical clustering analysis was performed using PermutMatrix (version 1.9.3, ATGC team, LIRMM, Montpellier, France) with the Pearson distance and Ward's aggregation method.

Statistical analysis of stress hormone data and lipid metabolomic data was performed using GraphPad Prism (version 7.03; GraphPad Software, Inc., San Diego, CA, USA). Data are expressed as the mean ± SEM. The significance of differences among the data were measured by one-way ANOVA followed by Tukey's post-hoc test, or by the Kruskal–Wallis test followed by Dunn's post-hoc test for data that did not follow the normal distribution.

#### *2.4. The Criteria for Metabolite Selection*

Metabolites were selected on the basis of the following criteria: a) all differences between groups were significant (*p* < 0.05), b) the metabolite level was at least twice as high in the AMD group than in the control group and similar between the control group and the AMD+PB group, and c) the highest relative level of the metabolite was greater than 10.

#### **3. Results and Discussion**

#### *3.1. Statistical Analysis of the Serum Metabolome*

Multivariate statistical analysis of the metabolome data was performed to identify statistically significant endogenous metabolites. First, PCA was conducted to determine the inherent similarities in the spectral profiles of the treatment groups. As shown in Figure 1, the control group and the AMD group were clearly divided into two clusters on the PCA score plot, whereas the AMD+PB group displayed a pattern almost identical to that of the control group. This result is in good agreement with the previous finding that the PB alleviated intestinal complications caused by AMD [19].

**Figure 1.** Principal component analysis (PCA) score plot of the metabolome analysis of the three treatment groups. Red triangles (-): anti-hypertensive medication amlodipine (AMD) group. Blue squares (): AMD plus a probiotic blend (AMD+PB) group. Green circles (-): control group.

#### *3.2. Screening and Identification of Candidate Markers for Lipid Metabolites*

To initially distinguish the differences among serum metabolites, hierarchical clustering analysis was performed to identify metabolites that were significantly increased or decreased among the treatment groups (Figure 2). Table 1 summarizes a detailed subgroup analysis of the metabolites. The G1a subgroup included cholesterol esters (CEs [22:6] and [20:4]), sterol derivatives, sphingomyelins (SMs), lysophosphatidylcholines (LysoPCs [18:0] and [16:0]), several phosphatidylcholines (PCs), and cholesterol. The G1b subgroup included PCs (18:0/22:6) and several triglycerides (TGs) with more than 10 double bonds (e.g., TG [60:12]). G1 metabolite levels overall were strongly increased in the AMD group but recovered in the AMD+PB group to the same level as those in the control group. On the other hand, the G2 metabolites included several TGs and diglycerides (DGs) with less than five double bonds (e.g., DG [34:1]). In addition, the TGs in G2 included monosaturated species (TG [48:1], TG [52:1], and TG [50:1]).

**Figure 2.** Hierarchical clustering analysis of the UPLC-HDMS metabolomics results. The rows display the metabolites, and the columns display the samples. Metabolites that significantly decreased relative to the average level across the samples are displayed in green, while those that significantly increased are displayed in red. The brightness of each color corresponds to the intensity of the difference compared with the average value.


**Table 1.** List of identified metabolites.

Abbreviations: C =null control group, A = AMD-administered group, A+P= AMD plus probiotic blend-administered group. Metabolites are arranged in order of the magnitude of the A/C fold change.

PCs and LysoPCs regulate immune function. PCs inhibit the TNF-α-induced upregulation of pro-inflammatory cytokines [22,23] and stimulate universal anti-inflammatory effects in the liver [24]. In contrast to the PCs, research on the immunomodulatory functions of LysoPCs shows conflicting results. Some studies show that LysoPCs contribute to the progression of inflammation by upregulating IL-1β-induced inducible nitric oxide synthase (NOS) [25] and also act as a death effector in the lipo-apoptosis of hepatocytes [26], which are key cells in innate immunity [27]. In addition, LysoPCs are involved in cardiovascular complications related to diabetes, rheumatoid arthritis, and atherosclerosis [28,29], as well as the activation of inflammatory responses via the acceleration of endothelial chemokine secretion [29,30]. However, other studies suggest that LysoPCs regulate inflammatory responses by inhibiting the secretion of pro-inflammatory cytokines such as TNF-α [31]. LysoPCs were evaluated as a biomarker because PC is converted to LysoPC by phospholipase A2 under inflammatory conditions [32–34]. Some studies suggest that LysoPCs are immunoregulatory lipid messengers under normal and pathogen-induced physiological conditions [35] because they can mediate signaling through G-protein-coupled receptors and be recognized as autoantigens [36]. Notwithstanding ambiguous results concerning pathways and mechanisms, it is certain that LysoPCs are involved in inflammation. Accordingly, it is apparent that increases of phospholipids such as PC and LysoPC in AMD-administered rats are associated with inflammation.

TGs are associated with the immune system. An excess of TGs causes diseases like hypertriglyceridemia [37], which is related to systemic inflammation [38]. Sterols play an essential role in countless biological processes including reproduction, metabolism, development, and immunity [39]. Cholesterols contribute to protection against infection by amplifying the inflammatory response and are the precursors of steroid hormones (including sex hormones, growth hormones, and glucocorticoids like corticosterone) [40]. However, excessive or prolonged cholesterol-induced immune responses can cause chronic inflammatory diseases like atherosclerosis [41]. Therefore, we performed additional analysis to examine and compare the levels of two hormones, adrenocorticotropic hormone (ACTH) and corticosterone, which are both representative stress hormones associated with immune reaction in rats [42–44].

#### *3.3. Identification and Comparison of Corticosterone and Adrenocorticotropic Hormone (ACTH)*

The probiotic blend used in this study was previously shown to have beneficial effects on human subjects with irritable bowel syndrome [45] and on animals with indomethacin-induced small intestine injury [46]. In our previous experiment, it was found to bring down increased levels of inflammatory cytokines in AMD-administered rats [19]. Glucocorticoids, including corticosterone in rodents and cortisol in humans, are anti-inflammatory steroid hormones [47,48]. In this context, we hypothesized that the probiotic blend could normalize potential anomalies in the level of corticosterone.

As shown in Figure 3, corticosterone levels were much higher in the AMD group than in the other two groups. The stress related to handling by the investigators was almost the same among the groups. Like the AMD group and the AMD+PB group, the control group was also subjected to oral gavage. Because the stress from oral gavage was the same among the groups, it is reasonable to hypothesize that higher corticosterone levels in the AMD group were caused by AMD-induced activation of the hypothalamic-pituitary-adrenocortical (HPA) axis or by a direct effect of AMD on the adrenal cortex, either or both of which were blocked by the PB co-treatment. In contrast, the ACTH levels decreased slightly more in the AMD group than in the control group (Figure 4). Considering that ACTH has a short half-life in plasma [49] and corticosterone itself is a negative regulator of ACTH secretion, the observed reduction of ACTH is likely a reflection of feedback inhibition of the HPA axis by corticosterone [47].

**Figure 3.** Corticosterone levels in the rat sera (\*; *p* value < 0.05, \*\*; *p* value < 0.01).

**Figure 4.** Adrenocorticotropic hormone (ACTH) levels in the rat sera.

#### **4. Conclusions**

Composition and stability of the gut microbiome is known to be affected by nutrition and disease, as well as antibiotics or medication [50]. Gut microbes influence lipid processing of hosts by engaging in gene expression related to the host's cholesterol and TG metabolism [51]. In this study, we revealed the change of lipid profiles in the serum of AMD-administered rats. Considering that impairment of the fine balance between gut microbes and the host's immune system leads to systemic inflammation [52], it can be postulated that the change of serum lipid profiles by AMD may reflect the disturbance of the gut microbial environment by AMD. Combined with these facts, our results suggest that AMD-induced dysbiosis leads to inflammation and changes in metabolic pathways, which in turn promotes the secretion of corticosterone to relieve the symptoms (Figure 5).

**Figure 5.** A cartoon summarizing the hypothesized effects of AMD-induced dysbiosis on lipid metabolism.

**Author Contributions:** Conceptualization, E.K. and Y.A.; methodology, M.H.N.; formal analysis, M.H.N.; investigation, Y.A.; resources, M.H.N. and Y.A.; data curation, Y.A. and M.H.N.; writing—original draft preparation, Y.A..; writing—review and editing, E.K.; visualization, Y.A.; supervision, E.K.; project administration, E.K.; funding acquisition, E.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was financially supported in part by Cell Biotech Co., Ltd. through the Cell Biotech R&D Centre and in part by the Strategic Initiative for Microbiomes in Agriculture and Food (918011-4) funded by the Ministry of Agriculture, Food and Rural Affairs. M.N. acknowledges the financial support from the National Research Foundation of Korea (NRF-2018M3A9F3056901).

**Acknowledgments:** Thanks to Kyeong-Hoon Jeong for advice on the stress hormones of mice.

**Conflicts of Interest:** The authors declare no conflicts 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* **The Timing E**ff**ects of Soy Protein Intake on Mice Gut Microbiota**

**Konomi Tamura 1,**†**, Hiroyuki Sasaki 1,2,**†**, Kazuto Shiga 1, Hiroki Miyakawa <sup>1</sup> and Shigenobu Shibata 1,\***


Received: 22 November 2019; Accepted: 25 December 2019; Published: 27 December 2019

**Abstract:** Soy protein intake is known to cause microbiota changes. While there are some reports about the effect of soy protein intake on gut microbiota and lipid metabolism, effective timing of soy protein intake has not been investigated. In this study, we examined the effect of soy protein intake timing on microbiota. Mice were fed twice a day, in the morning and evening, to compare the effect of soy protein intake in the morning with that in the evening. Mice were divided into three groups: mice fed only casein protein, mice fed soy protein in the morning, and mice fed soy protein in the evening under high-fat diet conditions. They were kept under the experimental condition for two weeks and were sacrificed afterward. We measured cecal pH and collected cecal contents and feces. Short-chain fatty acids (SCFAs) from cecal contents were measured by gas chromatography. The microbiota was analyzed by sequencing 16S rRNA genes from feces. Soy protein intake whether in the morning or evening led to a greater microbiota diversity and a decrease in cecal pH resulting from SCFA production compared to casein intake. In addition, these effects were relatively stronger by morning soy protein intake. Therefore, soy protein intake in the morning may have relatively stronger effects on microbiota than that in the evening.

**Keywords:** soy protein; microbiota; lipid metabolism; circadian; chrono-nutrition

#### **1. Introduction**

Mammals have approximately 100 trillion bacteria in their gut that comprise the microbiota. Gut microbiota has profound influences on the host's physiological conditions such as nutrient absorption, metabolism, and immunity [1]. Microbial alterations cause inflammatory bowel diseases and metabolic disorders. For example, concerning the microbiota of an obese person, the relative abundance of *Firmicutes*, which is a factor of obesity, is increased and *Bacteroidetes*, which prevents fat accumulation, is decreased [2]. In addition, an altered microbiota causes obesity because germ-free mice show an increase in body fat when injected with the microbiota of an obese mice [3].

Intestinal bacteria digest non-digestible food components such as dietary fibers, oligosaccharides, resistant starches, and resistant proteins, and produce short-chain fatty acids (SCFAs) [4,5]. SCFAs include acetic acid, propionic acid, lactic acid, and butyric acid. SCFAs are used as an energy source for colonic epithelial cells [6]. SCFAs maintain gut acidic conditions and prevent the growth of harmful bacteria such as *Enterobacteriaceae* and *Clostridia* [4,7]. SCFAs also have beneficial effects on mammalian energy metabolism and regulate the metabolism of fatty acid, glucose, and cholesterol [4].

The alteration of microbiota depends on various factors such as age, stress, disease, drugs, and diet [8]. There are many reports about the relationship between diet and microbiota. The microbiota can be rapidly affected by dietary changes [9]. Some studies evaluated the impact of protein on the microbiota. Many proteins are absorbed by the small intestine. However, segments of some proteins pass through the small intestine to reach the large intestine [10]. These resistant proteins and amino acids are metabolized by intestinal microbiota to SCFAs [11]. For example, soy protein intake causes higher microbial diversity and SCFA levels [12,13].

The circadian clock system plays an important role in maintaining physiological conditions such as the sleep-wake cycle, body temperature, and metabolism [14]. In the mammalian circadian system, there are two clocks: the main central oscillator in the suprachiasmatic nuclei (SCN), and the peripheral oscillator in peripheral organs. The SCN clock is mainly entrained by light-dark stimuli, and it regulates the peripheral clocks. The peripheral clocks are entrained by pharmacological agents, food nutrients, and mental or physical stress [15–19]. The SCFAs produced by the microbiota also entrain the circadian clock [20]. Microbiota exhibits diurnal oscillations in composition and function in both mice and humans. In addition, jet lag induced circadian disruption changes microbiota, and when feces from jet lag mice were transferred to germ-free mice, they became obese [21]. Thus, the activity of microbiota is strongly associated with circadian rhythm.

Chrono-nutrition is the science of nutrition, which is based on chronobiology. Hormonal secretion and the metabolism and absorption of nutrients have circadian variations [16]. Therefore, some food components have the most effective intake timing [22,23]. For example, fish oil intake in the morning rather than the evening is more effective to reduce lipids in mice [23]. Intake of water-soluble dietary fiber in the morning has a greater effect on microbiota diversity rather than in the evening [24]. In this study, we examined the effective timing of soy protein intake. Soy protein, especially β-conglycinin, is known to have beneficial effects on hepatic lipid metabolism, prevention of hepatic steatosis, and reduction of body fat in both rodents and humans [25–29]. It is also reported that soy protein intake has a superior effect on microbiota [12,26,27]. As hormonal secretion as well as microbial composition and function exhibit diurnal oscillations, the effective timing to alter microbiota can be different according to the food components [14,21]. However, the effective timing of soy protein intake has not been investigated. Therefore, in this study, we examined the effect of soy protein intake on mice gut microbiota based on chrono-nutrition, such as morning intake or evening intake.

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

#### *2.1. Animals and Diets*

We used 105 of ICR 8-week-old male mice (Tokyo Laboratory Animals, Tokyo, Japan) in this study. The mice were kept under 12 h light/12 h dark condition. Lights-on time was defined as zeitgeber time 0 (ZT0) and lights-off time as ZT12. Each mouse was housed in a plastic cage individually, at a temperature of 22 ± 2 ◦C, humidity of 60% ± 5%, and light intensity of 100-150 lux. We prepared two kinds of diet, a high-fat diet (HFD) with casein and HFD with soy protein (Fujipro F, Fuji Oil Co., Osaka, Japan) (Table 1).

To produce metabolic syndrome models with obesity, high inflammation, and abnormal microbiota, the mice were fed a HFD with casein and water ad libitum for one week before commencing the experiments. Thereafter, mice were fed HFD with casein or soy protein according to the experimental protocols. The Committee for Animal Experimentation at Waseda University approved all experimental protocols (permission protocol 2018-A030).


**Table 1.** Nutrition components (g) in each diet (100 g).

#### *2.2. Experimental Design*

In this study, we prepared two kinds of diets, HFD with casein (casein diet) and HFD with soy protein (soy diet), as described previously. For microbiota deterioration, mice were fed a casein diet for one week before commencing the experiments.

In experiment 1, we examined the effects of soy protein intake in a short period. Soy protein is known to have anti-obesity effects. Obesity causes a change in microbiota composition. Therefore, to eliminate the effect of body weight differences, we conducted the experiments over a short period, before body weight change. Mice were given free access to the casein diet (Casein group) or soy diet (Soy group) for 10 days, and they were sacrificed at ZT12, ZT20, or ZT4. Ten mice were prepared for each time point and group. We measured the cecal pH, and we collected the cecal contents, feces, blood, and liver samples (Figure 1a).

In experiment 2, we examined the timing effects of soy protein intake. To compare the effect of soy protein intake in the morning to that in the evening, mice were fed 1.8 g diets twice a day in the morning (ZT12) and evening (ZT20). When mice were given 4 h of access to food in the morning and evening, the amount of food consumption was different between the morning and evening [24]. Therefore, we fed mice 1.8 g diets twice a day, so that morning and evening food consumption is the same. It is reported that mice were able to consume all of 1.8 g diet within 4 h [30,31]. The experiment period was set to 14 days since it takes a few days for mice to adapt to the 2-meals-per-day feeding pattern. The mice were fed only casein diet (Casein group), soy diet in the morning and casein diet in the evening (M-Soy group), or casein diet in the morning and soy diet in the evening (E-Soy group). The mice were kept under experimental conditions for two weeks, and then they were sacrificed at ZT12, 20, or 4. Five mice were prepared for each time point and group. We measured the cecal pH and collected the cecal contents, feces, blood, and liver samples (Figure 1b).

#### *2.3. Cholesterol and Triglyceride Measurement*

Serum cholesterol and triglyceride (TG) levels were measured using cholesterol and triglyceride kit (FUJIFILM Wako Pure Chemical Co., Osaka, Japan). The assay was performed according to the manufacturer's instructions.

**Figure 1.** Experimental design. (**a**) Experimental protocol to examine the effect of soy protein intake. (**b**) Experimental protocol to examine the effect of soy protein intake timing. The white and black bars indicate environmental 12 h light and dark conditions, respectively. The horizontal blue arrow indicates free access to a high-fat diet (HFD) with casein. The horizontal orange arrow indicates free access to HFD with soy protein. The blue cylinder indicates the feeding timing of 1.8 g of HFD with casein. The orange cylinder indicates the feeding timing of 1.8 g of HFD with soy protein. The red triangles indicate the sampling time. High-fat diet, casein diet feeding, soy protein diet feeding, morning soy protein diet feeding, evening soy protein diet feeding: HFD, Casein, Soy, M-Soy, E-Soy, respectively.

#### *2.4. Real-Time RT-PCR*

Relative liver mRNA levels were measured by real-time RT-PCR. The mice were anesthetized with isoflurane and sacrificed at ZT 12, ZT 20, or ZT4. We collected livers to measure mRNA levels at each time point. Total liver RNA was extracted using RNA-*Solv* Reagent (Omega Bio-Tek Inc., Norcross, GA, USA). RNA concentration of each sample was adjusted using a spectrophotometer (GE Healthcare Japan Co., Tokyo, Japan). The RNA was reverse-transcribed and amplified using One-Step SYBR RT-PCR kit (Takara Bio Inc., Shiga, Japan) with specific primer pairs (Table 2) on Piko Real PCR system (Thermo Fisher Scientific, Waltham, MA, USA). The relative expression levels of target genes were normalized with *GAPDH*. The data were analyzed using the ΔΔCt method.


**Table 2.** Sequences of Primers for Real-time RT-PCR.

#### *2.5. Cecal pH Measurement*

Cecal pH was measured using pH meter (Euthech Instruments, Vernon Hills, IL, USA). The electrode of the pH meter was inserted directly into the cecum, immediately after collection.

#### *2.6. Short-Chain Fatty Acid (SCFA) Measurement*

Short-chain fatty acid (SCFA) in cecal contents was measured through gas chromatography (Shimadzu Co., Kyoto, Japan) as described in a previous report [32]. Cecal contents were acidified with sulfuric acid and SCFAs were extracted from 50 mg of cecal contents by shaking in 50 μL of sulfuric acid, 400 μL of diethyl ether, and 200 μL of ethanol (FUJIFILM Wako Pure Chemical Co., Osaka, Japan). The mixture was centrifuged at 18700× *g* for 30 s. The supernatant (1 μL) was injected into the capillary column (InertCap Pure WAX (30 m × 0.25 mm, df = 0.5 μm), GL Science, Tokyo, Japan) of gas chromatography coupled to a flame ionization detector. The initial temperature was 80 ◦C and the final temperature was 200 ◦C. Helium was used as carrier gas and quantification of the samples was performed using calibration curves for acetic, lactic, propionic, and butyric acids. A standard curve plotted for the quantitation of each acid was in the samples.

#### *2.7. Fecal DNA Extraction*

Fecal DNA was extracted according to the previous report with modifications [33]. We collected feces from the rectum, when we sacrificed the mice at each time point. Approximately 0.2 g fecal sample was suspended in a 50 mL tube containing 20 mL PBS. The suspension was filtered through a 100 μm nylon filter (Corning Inc., New York, NY, USA). The tube was washed with 10 mL PBS and then filtered through the filter. The filtrates were centrifuged at 9000× *g* for 20 min at 4 ◦C, and the supernatants were removed. Each precipitate was suspended in 1.5 mL TE 10 buffer (10 mM Tris-HCl (FUJIFILM Wako Pure Chemical Co., Osaka, Japan)/10 mM EDTA (DOJINDO, Tokyo, Japan)), and the suspension was transferred to 2 mL microtube. The suspensions were centrifuged at 9560× *g* for 5 min at 4 ◦C, and the supernatants were removed. Each precipitate was suspended in 800 μL TE 10 buffer. The suspensions were added 100 μL lysozyme (150 mg/mL) (FUJIFILM Wako Pure Chemical Co., Osaka, Japan) and then incubated for 1 h at 37 ◦C. Achromopeptidase (20 μL, 100 units/μL, FUJIFILM Wako Pure Chemical Co., Osaka, Japan) was added to the suspension and then incubated for 30 min at 37 ◦C. The suspension was treated with 50 μL of 20% sodium dodecyl sulfate and proteinase K (Promega Co., Madison, WI, USA) and then incubated for 1 h at 55 ◦C. To extract DNA, 980 μL PCI (phenol/chloroform/isoamyl alcohol) (Invitrogen, Carlsbad, CA, USA) was added and centrifuged at 6000× *g* for 10 min at 20 ◦C. The supernatant was transferred to a new 2 mL microtube and then suspended with 100 μL of 3 M sodium acetate and 900 μL isopropanol (FUJIFILM Wako Pure Chemical Co., Osaka, Japan). The suspensions were centrifuged at 6000× *g* for 10 min at 20 ◦C, and the supernatants were removed. The DNA pellet was rinsed with 1 mL of 70% ethanol and dried. The DNA was purified by treatment with 99 μL TE buffer and 1 μL RNase (10 μg/mL) (FUJIFILM Wako Pure Chemical Co., Osaka, Japan), then precipitated with 100 μL of 20% PEG solution (TOKYO Chemical Industry Co., Tokyo, Japan). The DNA was pelleted by centrifugation at 10,000× *g* at 4 ◦C, rinsed with 500 μL of 70% ethanol, and dissolved in 50 μL TE buffer.

#### *2.8. 16S rRNA Gene Sequencing*

16S rRNA gene sequencing was performed according to Illumina instructions (16S Metagenomic Sequencing Library Preparation). V3-V4 variable regions of the 16S rRNA gene were amplified using forward primer (5 -TCGTCGGCAGCGTCAGATG TGTATAAGAGACAGCCTACGGGNGGCWGCAG-3 ) and reverse primer (5 -GTCTCGTGGG CTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC-3 ). Amplicon PCR was performed with 2.5 μL microbial DNA (5 ng/μL), 5 μL of each amplicon PCR primer (1 μM), and 12.5 μL of 2 × KAPA HiFi HotStart Ready Mix (Kapa Biosystems, Wilmington, MA, USA) under conditions of 3 min at 95 ◦C, 25 cycles of 95 ◦C for 30 s, 55 ◦C for 30 s, and 72 ◦C for 30 s, and a final extension of 72 ◦C for 5 min. The PCR products were purified using AMPure XP beads (Beckman Coulter, Inc., Brea, CA, USA).

The Nextera XT Index Kit v2 (Illumina Inc., San Diego, CA, USA) was used to join dual indices and Illumina sequencing adapters. Index PCR was performed in 5 μL PCR production, 5 μL of each Nextera XT Index primer, 25 μL of 2 × KAPA HiFi HotStart Ready Mix, and 10 μL of PCR Grade water under conditions of 3 min at 95 ◦C, 8 cycles of 95 ◦C for 30 s, 55 ◦C for 30 s, and 72 ◦C for 30 s, and a final extension of 72 ◦C for 5 min. The PCR products were purified using AMPure XP beads. The quality of the purification was checked using Agilent 2100 Bioanalyzer with DNA 1000 kit (Agilent Technologies, Santa Clara, CA, USA). Finally, the DNA library concentration was diluted to 4 nM.

The DNA library was sequenced using Miseq reagent kit v3 (Illumina Inc., San Diego, CA, USA) on Illumina Miseq 2 × 300 bp platform. This sequencing was performed following manufacturer instructions.

#### *2.9. Analysis of 16S rRNA Gene Sequences*

16S rRNA gene sequence reads were processed through quantitative insights into microbial ecology (QIIME) pipeline version 1.9.1 [34]. Quality-filtered sequence reads were assigned to operational taxonomic units (OTUs) at 97% identity with the UCLUST algorithm [35]. These reads were then compared to the reference sequence collections in the Greengenes database (August 2013 version). A total of 4,034,110 reads were obtained from 105 samples. On average, 38,420 ± 2427 reads were obtained per sample. Taxonomy summary, alpha-diversity (within-sample), beta-diversity (between-sample dissimilarity), and principal coordinate analysis (PCoA) were calculated and generated by QIIME. PCoA analysis was also calculated using unweighted UniFrac distances.

#### *2.10. Metagenome Prediction*

The functional profiles of microbial communities were predicted through phylogenetic investigation of communities by reconstruction of unobserved states (PICRUSt) [36]. The functional predictions were assigned to almost all Kyoto Encyclopedia of Genes and Genomes (KEGG) ortholog (KO) functional profiles of microbial communities via 16S sequences. We selected and examined categories related to "Amino Acid Metabolism" and "Energy Metabolism" for analysis of simplification and clarity.

#### *2.11. Statistical Analysis*

Data are expressed as mean ± standard error of the mean (SEM). In this study, we compared the feeding condition at each time point, because we focused on the difference in feeding condition rather than time point. Statistical analysis was performed using GraphPad Prism version 6.03 (GraphPad Software, San Diego, CA, USA). The data were tested for normality and equality of variances using a D'Agostino-Pearson test/Kolmogorov-Smirnov test and Bartlett's test, respectively. Parametric analysis was conducted using one-way ANOVA with Tukey test or Student's *t*-test for post-hoc analysis, and non-parametric analysis was conducted using the Kruskal-Wallis test with Dunn's test or the Mann-Whitney test for post-hoc analysis. The differences in microbiota composition were tested using

the permutational multivariate analysis of variance (PERMANOVA). PERMANOVA was analyzed by QIIME.

#### **3. Results**

#### *3.1. Soy Protein Intake A*ff*ected Lipid Metabolism and the Gut Microbiota*

It has already been reported that soy protein not only reduces serum cholesterol and triglycerides, but also changes the microbiota composition, leading to considerable microbial diversity [12,25,26]. To examine whether the results of our study are similar to previous reports, we considered the effect of soy protein feeding in the free-feeding condition.

First, we examined the effect of soy protein on lipid metabolism. The food consumption (Casein group: 3.88 ± 0.15 g/day, Soy group: 3.92 ± 0.11 g/day) and final body weight (Casein group: 42.69 ± 0.59 g, Soy group: 42.15 ± 0.71 g) showed no differences between the groups. We showed the data of each time point and the average of a 3-time point (AVE). The serum cholesterol of the Soy group was significantly lower than that of the Casein group at ZT20 and AVE. Serum TG level showed no significant difference between both groups (Figure 2a). We measured the mRNA expression levels of fatty acid and cholesterol metabolism-related genes from liver samples. *Acc1*(ZT12, ZT20, and AVE), *Fasn* (ZT12, ZT20, and AVE), and *Srebp1c* (ZT12, ZT4, and AVE) expression levels were significantly lower in the Soy group than those in the Casein group. *Cyp7*α*1* expression level tended to be higher in the Soy group than that in the Casein group at ZT20 (Figure 2b).

To examine the effect of soy protein on microbiota, we measured cecal pH and SCFA production. Cecal pH was significantly lower in the Soy group than that in the Casein group (Figure 3a). Acetic acid (ZT12, ZT4, and AVE), propionic acid (ZT12 and ZT4), lactic acid, and butyric acid levels were significantly or tended to be higher in the Soy group than those in the Casein group (Figure 3b).

As cecal pH was decreased and SCFA production was increased, soy protein intake may alter the microbiota. Therefore, we analyzed the microbiota from feces. The Soy group showed significantly higher alpha-diversity for the Simpson index than that in the Casein group at ZT20 (Figure 4a). The PCoA of unweighted UniFrac distance showed that the beta-diversity of microbiota composition was significantly different between the Soy group and the Casein group (Figure 4b). Concerning the relative abundances of microbes at the phylum level, *Bacteroidetes* (ZT20) and *Proteobacteria* (ZT12, ZT20, and AVE) in the Soy group were significantly higher than those in the Casein group. *Firmicutes* (ZT12, ZT20, and AVE) in the Soy group were significantly lower than those in the Casein group (Figure 4c). At the genus level, *Bifidobacterium* (ZT12, ZT4, and AVE), *Enterococcus* (ZT20 and AVE), *[Ruminococcus]* (ZT20 and AVE), and *Desulfovibrio* (ZT20 and AVE) in the Soy group were significantly higher than those in the Casein group. *Lactococcus* in the Soy group was significantly lower than that in the Casein group (Figure 4d).

To infer the metagenome functional content based on the microbial community profiles obtained from 16S rRNA gene sequences, we used PICRUSt. The microbial communities could be distinguished based on their functions. The KEGG pathways associated with amino acid and energy metabolisms were significantly upregulated in the Soy group. The pathways associated with glycine, serine, and threonine metabolisms (ZT12, ZT20, and AVE) and lysine biosynthesis (ZT20 and AVE) in the Soy group were significantly upregulated compared to those in the Casein group (Figure 5a). The pathways associated with methane metabolism (ZT12, ZT20, and AVE) and nitrogen metabolism (ZT12, ZT20, and AVE) in the Soy group were significantly upregulated compared to those in the Casein group (Figure 5b).

**Figure 2.** Serum lipid and gene expression levels in the liver. (**a**) Serum cholesterol and triglyceride levels (ZT12, ZT20, ZT4, and an average of three points) of mice that were fed each diet for 10 days. (**b**) Relative RNA expression levels of fatty acid and cholesterol metabolism-related genes in the liver (ZT12, ZT20, ZT4, and an average of three points) of mice that were fed each diet for 10 days. Data are represented as mean ± SEM (*n* = 10). \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001 versus Casein, evaluated using Student's *t*-test. \$ *p* < 0.05, \$\$ *p* < 0.01, \$\$\$ *p* < 0.001 versus Casein, evaluated using the Mann-Whitney test. High-fat diet, casein diet feeding, soy protein diet feeding, an average value of three points: HFD, Casein, Soy, AVE, respectively.

**Figure 3.** The effect of soy protein intake on cecal pH and SCFA levels. (**a**) Cecal pH levels (ZT12, ZT20, ZT4, and an average of three points) of mice that were fed each diet for 10 days. (**b**) Cecal short-chain fatty acids (SCFA) levels (ZT12, ZT20, ZT4, and an average of three points) of mice that were fed each diet for 10 days. Data are represented as mean ± SEM (*n* = 10). \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001 versus Casein, evaluated using Student's *t*-test. \$ *p* < 0.05, \$\$ *p* < 0.01, \$\$\$ *p* < 0.001 versus Casein, evaluated using the Mann-Whitney test.

**Figure 4.** The effect of soy protein intake on microbiota. (**a**) Alpha-diversity about Simpson index (ZT12, ZT20, ZT4, and an average of three points) of mice that were fed each diet for 10 days. (**b**) Beta-diversity in comparison of each diet. The PCoA plots of unweighted UniFrac distance metrics obtained from sequencing the 16S rRNA gene in feces (*n* = 30). (**c**) The relative abundance of microbes at the Phylum level, and (**d**) at the Genus level of mice that were fed each diet for 10 days (ZT12, ZT20, ZT4, and an average of three points). Data (**a**,**c**,**d**) are represented as mean ± SEM (*n* = 10). \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001 versus Casein, evaluated using Student's *t*-test. \$ *p* < 0.05, \$\$ *p* < 0.01, \$\$\$ *p* < 0.001 versus Casein, evaluated using the Mann-Whitney test. The table in (**b**) indicates the result using PERMANOVA.

**Figure 5.** The functional predictions of microbial communities. (**a**) The functional predictions about categories related to "Amino Acid Metabolism" and (**b**) "Energy Metabolism" of microbial communities in mice that were fed each diet for 10 days (ZT12, ZT20, ZT4, and an average of three points). Data are represented as mean ± SEM (*n* = 10). \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001 versus Casein, evaluated using Student's *t*-test. \$\$ *p* < 0.01, \$\$\$ *p* < 0.001 versus Casein, evaluated using the Mann-Whitney test.

#### *3.2. Soy Protein Intake in the Morning A*ff*ected the Gut Microbiota More Than That in the Evening*

Since soy protein intake improved microbiota, we examined the effect of soy protein intake timing on microbiota. To compare the effect of soy protein intake in the morning and evening, mice were fed 1.8 g diets twice a day in the morning (ZT12) and evening (ZT20). Mice were fed only casein diet (Casein group), soy diet in the morning and casein diet in the evening (M-Soy group), or casein diet in the morning and soy diet in the evening (E-Soy group). Mice were kept under the experimental condition for two weeks and were then sacrificed at ZT12, 20 or 4 (Figure 1b). We measured cecal pH and collected cecal contents, feces, blood, and liver sample.

First, we examined the effect of soy protein intake timing on lipid metabolism. The final body weight (Casein group: 43.57 ± 0.76 g, M-Soy group: 43.86 ± 0.51 g, E-Soy group: 42.50 ± 0.77 g) showed no significant difference among the groups. We showed the data of each time point and the average of a 3-time point (AVE). Serum cholesterol in the M-Soy group was significantly higher than that in the Casein and the E-Soy groups at ZT12. The serum TG level showed no significant difference (Figure 6a). We measured the mRNA expression levels of fatty acid and cholesterol metabolism-related genes in the liver sample. *Acc1* expression level was significantly lower in the E-Soy group than that in the Casein group at ZT12 and AVE. *Fasn* expression level in the E-Soy group at ZT12 was significantly lower than that in the M-Soy group and tended to be lower than that in the Casein group. *Cyp7*α*1* expression level in the Casein group was significantly higher than that in the M-Soy group at ZT20 and ZT4, and that in the E-Soy group at ZT4 (Figure 6b).

**Figure 6.** Serum lipid levels and gene expression levels in the liver. (**a**) Serum cholesterol and triglyceride levels (ZT12, ZT20, ZT4, and an average of three points) of mice that were kept in each feeding condition for two weeks. (**b**) Relative RNA expression levels of fatty acid and cholesterol metabolism-related genes in the liver (ZT12, ZT20, ZT4, and an average of three points) of mice that were kept in each feeding condition for two weeks. Data are represented as mean ± SEM (*n* = 5). # *p* < 0.05, ## *p* < 0.01 evaluated using one-way ANOVA with Tukey's post-hoc test. % *p* < 0.05, %% *p* < 0.01 the Kruskal-Wallis test with Dunn's post-hoc test. High-fat diet, casein diet feeding, morning soy protein diet feeding, evening soy protein diet feeding, the average value of three points: HFD, Casein, M-Soy, E-Soy, AVE, respectively.

To examine the effect of soy protein intake timing on microbiota, we measured cecal pH and SCFA production. Cecal pH in the M-Soy group tended to be lower than that in the Casein group at ZT12, and significantly lower than that in the other groups at ZT20. The E-Soy group showed a significantly lower pH than those in the other groups at ZT4. The M-Soy and the E-Soy groups showed significantly lower cecal pH than that in the Casein group on AVE (Figure 7a). The lactic acid in the M-Soy group was significantly higher than that in the Casein group and tended to be higher than that in the E-Soy group at ZT20. The lactic acid in the E-Soy group was significantly higher than that in the Casein group at ZT4. Only the M-Soy group showed a significantly higher level of lactic acid than that in the Casein group on AVE. Butyric acid in the M-Soy group tended to be higher than that in the Casein group at ZT20 and ZT4, and significantly higher than that in the E-Soy group at ZT20. Butyric acid in the E-Soy group was significantly higher than that in the other groups at ZT4. Only the M-Soy group showed a significantly higher level of butyric acid than that in the Casein group on AVE (Figure 7b).

**Figure 7.** The effect of soy protein intake timing on cecal pH and SCFA levels. (**a**) Cecal pH levels (ZT12, ZT20, ZT4, and an average of three points) of mice that were kept in each feeding condition for two weeks. (**b**) Cecal SCFA levels (ZT12, ZT20, ZT4 and average of three points) of mice that were kept in each feeding condition for two weeks. Data are represented as mean ± SEM (*n* = 5). # *p* < 0.05, ## *p* < 0.01, ### *p* < 0.001 evaluated using one-way ANOVA with Tukey's post-hoc test. % *p* < 0.05, %% *p* < 0.01 the Kruskal-Wallis test with Dunn's post-hoc test.

As the cecal pH was decreased and SCFA production was increased, soy protein intake in the morning may strongly alter the microbiota. Therefore, we analyzed the microbiota from feces. The M-Soy group showed tend to higher alpha-diversity for Simpson index than that shown by the other groups on AVE (Figure 8a). The PCoA of unweighted UniFrac distance showed that the beta-diversity of microbiota composition was significantly different between the Casein and the M-Soy group at ZT20. On the other hand, between the Casein and the E-Soy group, the beta-diversity of microbiota composition didn't show a significant difference at ZT4 (Figure 8b). The PCoA of unweighted UniFrac distance of all-time points was significantly different between the Casein and the M-Soy groups (statistic value = 2.478, *p* = 0.002), and relatively different between the Casein and the E-Soy groups (statistic value = 1.460, *p* = 0.048) (Figure 8c). For the relative abundance of microbes at the phylum level, *Bacteroidetes* was significantly higher in the M-Soy group than that in the Casein group on AVE. The relative abundance of *Firmicutes* was significantly lower in the E-Soy group than that in the Casein group at ZT20 and AVE (Figure 8d). In the genus level, the relative abundance of *Lactococcus* in the M-Soy group was significantly lower than that in the other groups at ZT20, and the relative abundance of *Lactococcus* in E-Soy group was significantly lower than that in the other groups at ZT4. On AVE, the relative abundance of *Lactococcus* in the M-Soy group was significantly lower and that in the E-Soy group tended to be lower than that in the Casein group. The relative abundance of *[Ruminococcus]* in the E-Soy group was significantly lower than that in the other groups at ZT20 (Figure 8e).

To infer the metagenome functional content based on microbial community profiles obtained from 16S rRNA gene sequences, we used PICRUSt. The microbial communities could be distinguished based on their functions. The KEGG pathways associated with glycine, serine, and threonine metabolism in the M-Soy group were significantly or tended to be upregulated at ZT12 and AVE compared to those in the Casein group (Figure 9a). The pathways associated with methane metabolism in the M-Soy group were significantly or tended to be upregulated compared to those in the Casein group at ZT12 and AVE, and compared to those in the E-Soy group at ZT20. The pathways associated with methane metabolism in the E-Soy group were significantly upregulated compared to those in the Casein group at ZT12. The pathways associated with nitrogen metabolism in the M-Soy group tended to be upregulated compared to those in the other groups on AVE (Figure 9b).

**Figure 8.** The effect of soy protein intake timing on microbiota. (**a**) Alpha-diversity about Simpson index (ZT12, ZT20, ZT4, and an average of three points) of mice that were kept in each feeding condition for two weeks. (**b**) Beta-diversity in comparison of Casein and M-Soy at ZT20, or Casein and E-Soy at ZT4, shortly after soy protein intake (*n* = 5). (**c**) Beta-diversity in comparison of Casein and M-Soy, Casein and E-Soy, or M-Soy and E-Soy. The PCoA plots of unweighted UniFrac distance metrics obtained from sequencing the 16S rRNA gene in feces (*n* = 15). (**d**) The relative abundance of microbes at the Phylum level, and (**e**) at the Genus level of mice that were kept in each feeding condition for two weeks. Data (**a**,**d**,**e**) are represented as mean ± SEM (*n* = 5). # *p* < 0.05 evaluated using one-way ANOVA with Tukey's post-hoc test. % *p* < 0.05, %% *p* < 0.01 the Kruskal-Wallis test with Dunn's post-hoc test. The tables in (**b**,**c**) indicate the result using PERMANOVA.

**Figure 9.** The functional predictions of microbial communities. (**a**) The functional predictions about categories related to "Amino Acid Metabolism" and (**b**) "Energy Metabolism" of microbial communities in mice that were kept under each feeding condition for two weeks (ZT12, ZT20, ZT4, and an average of three points). Data are represented as mean ± SEM (*n* = 5). # *p* < 0.05, ## *p* < 0.01 evaluated using one-way ANOVA with Tukey's post-hoc test.

#### **4. Discussion**

In this study, 10 days of soy protein intake reduced serum cholesterol and fatty acid synthesis related genes expression levels were experimented with in mice. In addition, soy protein changed microbial conditions and decreased cecal pH caused by SCFA production. Two weeks of soy protein feeding in the morning or evening resulted in a decrease in cecal pH and an increase in SCFA and microbiota diversity change after soy protein intake. In addition, soy protein intake in the morning may have a longer effect on SCFA production and cecal pH reduction than that of soy protein intake in the evening. It was suggested that soy protein might attenuate abnormality in gut microbiota effectively when taken in the morning rather than in the evening.

In experiment 1, soy protein reduced serum cholesterol level and fatty acid synthesis related genes such as *Acc1*, *Fasn*, and *Srebp1c* expression levels (Figure 2). It has already been reported that soy protein reduces serum cholesterol, TG, and fatty acid synthesis related genes expression levels [12,26]. In these reports, the effects of long-term soy protein intake were examined. In our study, mice were fed soy protein only for 10 days. However, serum cholesterol and fatty acid synthesis related genes expression levels were decreased. It was reported that SCFAs produced by the microbiota upregulated the expression of GLP-1 via activation of the MAPK pathway. GLP-1 induced reduction in mRNA expression of the fatty acid synthesis related genes [37–39]. Therefore, it is suggested that SCFA production by soy protein intake may be related to a reduction in fatty acid synthesis related gene expression. Since the current study focused on the effects of soy protein on microbiota, we measured

only mRNA levels but not protein levels of the fatty acid synthesis related genes. To support these mRNA data, it might be necessary to measure the protein levels of enzyme activity.

Soy protein also changed the microbiota composition (Figure 4). Soy protein intake enhanced the production of SCFA, especially lactic and butyric acids, and decreased cecal pH (Figure 3). Previous studies also reported that soy protein causes greater diversity of microbiota than milk protein does, and their microbiota showed different compositions [12,25–27]. However, these previous studies examined the effects of long-term soy protein intake with bodyweight changes. Therefore, these reports could not exclude the possibility that soy protein provides anti-obesity effects and then improves the microbiota. In the current study, only 10 days of soy protein intake changed the microbiota without body weight changes. Our previous study reported that water-soluble dietary fiber changed microbiota in 10 days [24]. It is suggested that soy protein intake itself changes the microbiota and even short-term intake is effective in changing the microbiota.

In the present study, the relative abundances of various bacteria were changed by soy protein intake both at the phylum and genus levels (Figure 4c,d). These results were similar to previous reports [27,40]. *Firmicutes* are known as the obese factor and its relative abundance is higher in obese people [2]. In this experiment, *Firmicutes* was decreased by soy protein intake. This result suggests that soy protein may have an anti-obesity effect by decreasing *Firmicutes*. *Bifidobacterium* and *Enterococcus* are known to produce acetic and lactic acids through fermentative metabolism [41–43]. An increase in the relative abundance of *Bifidobacterium* and *Enterococcus* might cause an increase in acetic and lactic acids. It has also been reported that oral administration of *Bifidobacterium breve* to infants may prevent digestive disease [44]. Thus, Soy protein intake may prevent diseases of the intestines by increasing the relative abundance of *Bifidobacterium*. These changes in microbiota may be related to a decrease in serum lipid and hepatic fatty acid synthesis related gene expression. *Lactococcus* is known to produce lactic acid. However, *Lactococcus* was decreased by soy protein intake in this study. The reason may be because *Lactococcus* is the bacteria commonly found in raw milk, cheese, and other dairy products [45]. In this experiment, *Proteobacteria* and *Desulfovibrio* were increased by soy protein intake. It was reported that an increase in the abundance of *Proteobacteria* can change microbiota [46]. Further, *Desulfovibrio* in *Proteobacteria* can induce barrier dysfunction [47]. In the present study, we observed a greater diversity in the Soy group. However, some negative bacteria were increased. The reason underlying this increase in bacterial number and the associated mechanism under soy protein feeding condition remain unclear. In the future, this may be clarified by examining in detail the relationship between diet and microbiota.

In experiment 2, we examined the effect of soy protein intake timing (morning or evening) on microbiota. There were smaller effects of the diets on serum lipid, hepatic fatty acid, and cholesterol metabolism-related gene expression levels in experiment 2 than those in experiment 1 ( Figure 2; Figure 6), because the amount of soy protein intake was smaller in experiment 2 than in experiment 1. In addition, it has been reported that mice under time-restricted feeding of 8 h per day were protected against obesity and hepatic steatosis, with improved energy expenditure [48]. In this experiment, mice were restricted in not only the amount but also the timing of feeding. Therefore, the feeding schedule itself may have powerfully reduced the serum lipid levels and hepatic gene expression of the fatty acid synthesis related genes, and then this protocol may mask the effects of soy protein.

At first, we compared the effects of soy protein on the microbiota at 8 h after feeding initiation in both the groups (ZT20 for the morning intake and ZT4 for the evening intake), because microbiota was altered shortly after water-soluble dietary fiber intake under two-meals-per-day schedule [24] and rapidly affected by dietary changes [9]. The cecal pH and the amount of lactic acid and butyric acid showed similar effects in both morning and evening groups (Figure 7). These data suggest that soy protein has beneficial effects on the microbiota in the morning intake or evening intake as compared to the casein intake. In addition, the previous feeding has a strong effect on microbial conditions. Since we collected samples every 8 h in this study, we can discuss the effects throughout the day by taking the average of 3-time points. The morning soy protein intake also showed lower cecal pH before soy protein intake (ZT12) and higher levels of lactic and butyric acids at a one day average

(Figure 7). Overall, it is suggested that soy protein intake in the morning may have long term effects on SCFA production and cecal pH reduction than that of the soy protein intake in the evening. Regarding microbiota, soy protein intake in the morning, not in the evening caused greater diversity on one day average (Figure 8a). In addition, microbiota varied to a greater extent by soy protein intake in the morning than that in the evening, shortly after soy protein intake (Figure 8b), and throughout the day (Figure 8c).

We do not know the detailed mechanism of such different effects on microbiota between morning intake and evening intake groups. The difference between soy protein intake in the morning and evening may be caused by a difference in fasting time before each diet. Two-meal-per-day feeding conditions were set close to a human feeding pattern in this study. In general, the fasting period before breakfast is the longest compared to other meal times in human eating habits. It has been reported that food signal after a long fasting strongly determines the peripheral clock phase [49]. It is also reported that consumption of water-soluble dietary fiber at breakfast which is after a longer fasting period, had greater effects on the microbiota [24]. Soy proteins contain proteins that are resistant to digestion [50], therefore resistant protein becomes the good food for microbiota just like water-soluble dietary fiber. In this study, fasting time was longer before the morning diet than before the evening diet. The intake of soy protein including resistant protein after long fasting may also have a greater impact on the microbiota. Furthermore, it has been reported that microbiota composition has circadian dynamics [21]. Even in mice that were fed the same diet, different reactions may occur if the microbiota is different in the morning or evening. The abdominal temperature, bowel movement, and endocrine system may influence the microbiota diversity [51–54], and these factors show circadian rhythm [55–57]. Thus, the different microbial reactions that were observed based on the soy protein intake in the morning or evening may be explained by the differences in the fasting time before each diet, microbiota circadian oscillations, and gut functional rhythm.

We used PICRUSt analysis to infer the functional capabilities of microbial communities. Soy protein intake upregulated the KEGG pathways associated with amino acid metabolism, especially glycine, serine, and threonine metabolism and lysine biosynthesis (Figure 5a). It has been reported that downregulating the pathway associated with amino acid metabolism has been observed in diarrheic calves and dogs and may be a feature of microbiota-associated diseases [58,59]. It is suggested that soy protein may improve microbiota. The pathways associated with methane and nitrogen metabolisms in the Soy group were significantly upregulated compared to those in the Casein group (Figure 5b). The upregulation of the nitrogen metabolism pathway by soy protein intake suggested that the indigestible component of soy protein might be metabolized by microbiota. The pathways associated with glycine, serine, and threonine metabolism, methane metabolism, and nitrogen metabolism were significantly upregulated in the M-Soy group compared to those in the Casein group (Figure 9). It is suggested that soy protein intake in the morning may have a stronger effect on upregulating these pathways than that in the evening. However, PICRUSt is only a predictor of metagenomic function. Therefore, metabolomic approaches are preferred in identifying factual changes in the metabolic function of microbiota by soy protein intake and its timing and identifying biomarkers for unstable gut microbiota.

#### **5. Conclusions**

In summary, the present experiments showed that soy protein intake and its timing affected the microbiota. The change in microbiota caused SCFA production and a decrease in cecal pH. In particular, soy protein may be effective in improving lipid metabolism and changing microbiota even with short-term intake. In addition, with respect to the timing of soy protein intake, morning intake may have relatively stronger effects on microbiota than evening intake would. This study provides evidence that soy protein intake and its timing are important factors that affect microbiota composition. To our knowledge, this is the first study to examine the effect of protein intake timing on microbiota and predict the functional profiles of microbial communities affected by soy protein. Therefore, our results

are expected to be useful in designing future studies that may focus on the effects of foods or beverages in improving microbiota composition at different mealtimes and in providing important information for chrono-nutrition research.

**Author Contributions:** Research design and data analysis, K.T., H.S., S.S.; writing manuscript, K.T., H.S., S.S.; performing experiment, K.T., H.S., K.S., H.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was partially supported by the Council for Science, Technology, and Innovation, SIP, "Technologies for creating next-generation agriculture, forestry, and fisheries" (funding agency: Bio-oriented Technology Research Advancement Institution, NARO) (Shibata. S.), Japan Society for the Promotion of Science (JSPS) KAKENHI (A and Houga) (Shibata. S), and Fuji Foundation for Protein Research (Sasaki. H).

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

#### **References**


© 2019 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*

### **Urinary TMAO Levels Are Associated with the Taxonomic Composition of the Gut Microbiota and with the Choline TMA-Lyase Gene (***cutC***) Harbored by Enterobacteriaceae**

**Alessandro Dalla Via 1, Giorgio Gargari 1, Valentina Taverniti 1, Greta Rondini 1, Ilaria Velardi 1, Veniero Gambaro 2, Giacomo Luca Visconti 2, Valerio De Vitis 1, Claudio Gardana 1, Enzio Ragg 1, Andrea Pinto 1, Patrizia Riso <sup>1</sup> and Simone Guglielmetti 2,\***


Received: 7 December 2019; Accepted: 19 December 2019; Published: 25 December 2019

**Abstract:** Gut microbiota metabolization of dietary choline may promote atherosclerosis through trimethylamine (TMA), which is rapidly absorbed and converted in the liver to proatherogenic trimethylamine-N-oxide (TMAO). The aim of this study was to verify whether TMAO urinary levels may be associated with the fecal relative abundance of specific bacterial taxa and the bacterial choline TMA-lyase gene *cutC*. The analysis of sequences available in GenBank grouped the *cutC* gene into two main clusters, cut-Dd and cut-Kp. A quantitative real-time polymerase chain reaction (qPCR) protocol was developed to quantify *cutC* and was used with DNA isolated from three fecal samples collected weekly over the course of three consecutive weeks from 16 healthy adults. The same DNA was used for 16S rRNA gene profiling. Concomitantly, urine was used to quantify TMAO by ultra-performance liquid chromatography coupled with tandem mass spectrometry (UPLC-MS/MS). All samples were positive for *cutC* and TMAO. Correlation analysis showed that the cut-Kp gene cluster was significantly associated with *Enterobacteriaceae*. Linear mixed models revealed that urinary TMAO levels may be predicted by fecal cut-Kp and by 23 operational taxonomic units (OTUs). Most of the OTUs significantly associated with TMAO were also significantly associated with cut-Kp, confirming the possible relationship between these two factors. In conclusion, this preliminary method-development study suggests the existence of a relationship between TMAO excreted in urine, specific fecal bacterial OTUs, and a *cutC* subgroup ascribable to the choline-TMA conversion enzymes of *Enterobacteriaceae*.

**Keywords:** choline; trimethylamine; trimethylamine n-oxide; 16S rRNA gene profiling; qPCR; linear mixed models

#### **1. Introduction**

From infancy, the microorganisms colonizing the human gastrointestinal tract (GIT), collectively known as GIT microbiota, act as a "hidden" metabolic organ that exerts indispensable functions for the development and physiology of the human organism, such as the production of vitamins, modulation of the immune system, competitive exclusion toward exogenous pathogenic bacteria,

xenobiotic detoxification, and production of short-chain fatty acids [1]. Nonetheless, detrimental activities have also been associated with gut commensal microorganisms, such as the production of carcinogens by the bacterial nitroreductases and azoreductases [2], or the conversion of primary bile acids to toxic compounds by the microbiota-associated enzyme cholesterol dehydrogenase and 7-α-dehydroxylase [3]. In addition, it was proposed that the intestinal bacterial enzymatic activities that produce trimethylamine (TMA) may promote atherosclerosis. TMA, in fact, is readily absorbed from the intestinal tract and, once in the liver, is converted into trimethylamine-N-oxide (TMAO) [4], whose plasma level has been identified as a metabolite strongly associated with atherosclerosis in a large case-control cohort for cardiovascular disease [5]. In particular, TMAO was proposed to promote atherogenesis by increasing cholesterol in macrophages and enhancing the accumulation of foam cells in artery walls [4,5]. Nonetheless, the literature has contradicted the role of TMAO, and recent studies have questioned its deleterious role in the cardiovascular system [6], suggesting, on the contrary, that TMAO could have protective functions [7,8].

Reportedly, a dominant contribution to the production of TMA in the gut comes from the microbial metabolism of diet-derived substrates such as carnitine- and choline-containing molecules [4,5,9]. Choline is an essential nutrient that is used by cells to synthesize membrane phospholipids. Furthermore, choline is the precursor of the neurotransmitter acetylcholine and a major source for methyl groups via its metabolite, trimethylglycine (betaine) [10]. The main dietary sources of the choline moiety, which is mostly present in food as lecithin (i.e., phosphatidylcholine), were reported to be eggs, liver, soybeans, and pork [11]. Although they are also present in numerous other foods [12], recent surveys in the USA indicated that choline may be underconsumed in specific populations (e.g., pregnant women and vegans) [13]. Based on the average observed choline intake in healthy European populations, a panel of the European Food Safety Authority set the adequate intake of choline at 400 mg/day [14].

Recent literature has suggested that the enhanced abundance of choline utilization genes in the intestinal microbiome is associated with increased TMA levels in the gut and, subsequently, with a higher hepatic production of TMAO. Proof of the importance of choline-derived TMA in the context of TMAO toxicity was recently provided by the study of Craciun and Balskus, in which the specific inhibition in mouse intestine of the microbial choline TMA-lyase (the primary enzymatic activity involved in the production of TMA from choline [15]) resulted in a significant reduction in plasma TMAO levels and recovery from dietary-induced platelet aggregation and thrombus formation [16].

Choline TMA-lyase is discontinuously distributed in bacterial taxa. Consequently, it was speculated that the phylogenetic composition of the microbiota is plausibly a poor predictor of the intestinal potential to convert choline into TMA [15,17,18]. However, in another study, the taxonomic structure of the gut microbiota was used to predict genes involved in choline metabolism [19] by means of PICRUSt, a bioinformatic tool used to infer the functional profiles of the microbial communities from 16S rRNA gene profiling data [20]. Although the toxicity of TMAO has been extensively investigated in the last 10 years, the association potentially existing among host TMAO levels, gut microbiota composition, and the intestinal microbial metabolization of choline has been only marginally considered. In this context, we developed a molecular protocol for the targeted quantification in the fecal microbiome of the bacterial gene *cutC* coding for the glycyl radical enzyme homolog choline TMA-lyase [15,21]. This protocol was applied to quantify the *cutC* gene abundance in the fecal samples collected at different time points from a group of healthy adults. Then, the obtained results were analyzed in comparison with the bacterial taxonomic composition and the urinary levels of TMAO concomitantly determined in the same population to deduce the potential association of excreted TMAO with gut microbial taxa and/or specific choline TMA-lyase enzymes.

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

#### *2.1. Design and Use of Primers Targeting the* cutC *Gene*

The primers used in polymerase chain reaction (PCR) for the amplification of the *cutC* gene were designed as follows. The GenBank database and Conserved Domain Database (CDD) at the National Center for Biotechnology Information (NCBI) were queried to select 52 nonredundant representative bacterial proteins of the choline trimethylamine-lyase protein family TIGR04394 (choline\_CutC; EC Number 4.3.99.4), including the CutC enzymes of *Desulfovibrio desulfuricans* [4], and *Klebsiella pneumoniae* [22]. Then, the corresponding CDS nucleotide sequences of selected proteins were used to build a UPMGA tree upon ClustalW multiple alignments. According to the obtained dendrogram, sequences were clustered in two groups: One including the *cutC* sequence of *K. pneumoniae*, named cut-Kp, and one including the *cutC* sequence of *D. desulfuricans*, named cut-Dd (Supplementary Figure S1). Finally, a pair of primers was designed in the most conserved regions of each group of sequences: cut-Dd-F, 5 -CGTGTTGACCAGTACATGTA-3 and cut-Dd-R 5 -GCTGGTAACCTGCGAAGAA-3 (expected amplicon of 185 bp); cut-Kp-F, 5 -GATCTGACCTATCTGATTATGG-3 , and cut-Kp-R, 5 -TTGTGGAGCATCATCTTGAT-3 (expected amplicon of 190 bp).

#### *2.2. PCR Detection of* cutC *Gene in Single Strains*

The two primer pairs designed as described above were used in endpoint PCR with the genomic DNA extracted from 64 bacterial strains (Table S1). Reaction mix was prepared in 25 μL, including 0.5 units of DreamTaq Polymerase (ThermoFisher, Fermentas, Waltham, MA, USA), 1× concentration of DreamTaq Polymerase Buffer (ThermoFisher, Fermentas,), 0.25 μM of each primer, 200 μM of deoxyribonucleotide triphosphate (dNTPs), and 0.5 mM of MgCl2. The PCR cycle program used was the following: Initial denaturation at 95 ◦C for 2 min, followed by 35 cycles of denaturation at 94 ◦C for 45 s, annealing at 58 ◦C for 45 s for the cut-Dd couple and 56 ◦C for 45 s for the cut-Kp couple, and extension at 72 ◦C for 20 s. A final extension of 7 min at 72 ◦C was then applied.

#### *2.3. Detection of Choline-Utilization Activity in Single Strains*

Bacterial strains were grown in the respective culture medium (reported in Table S1) for 48 h. Afterward, the biomasses were collected by centrifugation at 9500 g for 10 min. The cell pellets were then washed with sterile PBS and resuspended in fresh medium with the addition of 0.2% filter-sterilized choline. Bacteria were incubated at 37 ◦C for 48 h in glass tubes with screw cap. Afterward, supernatants were collected and used for mass spectrometry (MS) and nuclear magnetic resonance (NMR) analyses. The MS analyzes were performed by directly injecting 5 μL of diluted broth cultures after the removal of the bacterial cells by centrifugation and subsequent filtration with a 0.45-μm syringe filter. In detail, the broth cultures were analyzed in full scan in the range from 50 u to 400 u on an HR-MS Orbitrap model Exactive with a HESI-II probe for electrospray ionization (Thermo Scientific, San Jose, CA, USA). The resolution, gain control, mass tolerance, and maximum ion injection time was set to 50 K, 1E6, 2 ppm, and 100 ms, respectively. The MS data were processed using Xcalibur software (Thermo Scientific). Choline and TMA were used as reference standard. Choline and TMA were also directly detected in broth cultures by 1H-NMR with a 60 MHz benchtop NMR spectrometer Spinsolve 60 Carbon Ultra, Magritek GmbH (Aachen, Germany).

#### *2.4. Study Population*

Study participants were recruited within the University campus. In total, four females and 12 males aged 21–45 (mean: 29.8 years) were enrolled (Table S2). The inclusion criteria were as follows: Healthy adult volunteers of both sexes who provided signed informed consent of their participation in the study. The exclusion criteria were as follows: Antibiotic consumption in the month preceding the start of the study, consumption of antacids or prokinetic gastrointestinal drugs, episodes of viral or bacterial enteritis in the two months prior to the study, episodes of gastric or duodenal ulcers in the

previous five years, pregnancy or breastfeeding, recent history of alcohol abuse or suspected drug use, and any severe disease that may interfere with treatment. Ethical permission was granted by the University of Milan Ethics Committee (ref: opinion no. 37/16, 15 December 2016).

#### *2.5. Collection of Fecal and Urine Samples*

Three fecal sample were collected weekly over the course of three consecutive weeks from each volunteer. All the participants were asked to follow their regular diet during the three weeks. Concomitantly to the fecal sample, the volunteers provided 24-h urine collection.

Urine samples were collected over 24 h in sterile tanks and on the same days that fecal samples were been collected. The volume of collected urine was recorded in order to calculate the daily excretion of trimethylamine oxide (TMAO). Immediately after delivery, part of the urine samples was transferred in 10-mL sterile tubes and stored at −80 ◦C until analysis.

#### *2.6. Analysis of* cutC *Gene by Quantitative Real-Time PCR*

The cutC gene was quantified in fecal DNA with quantitative real-time PCR (qPCR) with both primer pairs, cut-Dd and cut-Kp. To this aim, DNA was extracted from feces using the kit PowerLyzer® PowerFecal® DNA Isolation Kit (MO BIO Laboratories, Inc.), starting from 0.25 <sup>±</sup> 0.02 mg of sample according to the manufacturer's instructions. Primer pairs were tested with a gradient qPCR in a range of eight temperatures in order to find the most efficient annealing temperature using DNA of Streptococcus dysgalactiae 485 and Klebsiella sp. A1.2 as reference DNA. In addition, the amplification efficiency of the two pairs of primers was tested in qPCR experiments with six serial 1:3 dilutions of genomic DNA isolated from *Streptococcus dysgalactiae* 485, *Klebsiella* sp. A1.2, and human fecal metagenomic DNA. All DNA (bacterial and metagenomic) serial dilutions were tested with primer concentrations of 0.5 μM, 0.4 μM, and 0.3 μM. Efficiency curves were obtained with Bio-Rad software by setting samples as "standard" and obtaining a curve with efficiency (E) parameter and R2 value. Based on the results of these setup experiments, primers were then used at a final concentration of 0.5 μM, as with this concentration, we obtained an R2 value of 0.98. In addition, two randomly selected fecal DNA samples were tested at the different concentration by adding 70 ng, 50 ng, 25 ng, and 10 ng in qPCR reactions. Based on Ct value comparison between the different DNA concentrations, the *cutC* gene quantification was subsequently performed using 50 ng of total DNA. The reaction mix contained the SsoFast TM Eva-SuperGreen Supermix 2× (Bio-Rad Laboratories), deionized Milli-Q water (Millipore), and primers. All DNA samples (5 μL in each well) were tested in technical duplicate. The qPCR cycles employed were the following: Initial denaturation at 95 ◦C for 3 min, followed by 44 cycles of denaturation at 95 ◦C for 30 s, annealing at 58 ◦C (for cut-Dd primers) or 58.5 ◦C (for cut-Kp primers) for 30 s, and elongation at 72 ◦C for 5 s. A final denaturation ramp between 65 ◦C and 95 ◦C for 5 s was performed for the melting curve analysis. Moreover, specificity of qPCR reaction was confirmed by checking the presence of only one amplification and of the expected size in electrophoresis on a 2% agarose gel. A total of 48 fecal samples were analyzed. Each sample was analyzed with each primer set in duplicate. The 2−ΔΔCt method was used for the relative quantification of *cutC* gene, using the EUB panbacterial primers [Muyzer] targeting the 16S rRNA gene as reference. Data were reported as relative increase of *cutC* copy number compared to the level of the sample that showed the highest significant Ct in qPCR set as 1.

#### *2.7. Analysis of the Bacterial Taxonomic Composition of Fecal Samples*

The bacterial community structure of the fecal microbiota was analyzed as described elsewhere [23,24], with DNA extracted from feces as described in Section 2.2. In brief, extracted DNA was analyzed through 16S rRNA gene profiling. Sequencing reads were generated at the Institute for Genome Sciences (University of Maryland, School of Medicine, Baltimore, MD, USA) with Illumina HiSeq 2500 rapid run sequencing of the V3–V4 variable region. Sequencing reads were equally distributed among the samples. Sequences were filtered and trimmed based on their quality. We obtained a sequence length of 301 bp for both R1 and R2 sequences with an average quality score (Phred score) higher than 35. Sequencing reads were rarefied at 5000 per sample. Subsequently, sequence reads were analyzed through the bioinformatic pipeline Quantitative Insights into Microbial Ecology (QIIME) version 1.9.1 [25] with the GreenGenes database updated to version 13.5. The relative abundance of bacteria in each fecal sample was reported at the taxonomic levels of phylum, class, order, family, genus, and operational taxonomic units (OTUs). Sequence were deposited in the European Nucleotide Archive (ENA) of the European Bioinformatics Institute under accession code PRJEB34169.

#### *2.8. TMAO Quantification in Urine Samples*

TMAO levels in urine samples were determined by ultra-performance liquid chromatography coupled to tandem mass spectrometry (UPLC-MS/MS) (Waters Acquity UPLC system). The analysis method involved the use of a totally porous column with stationary C8 stable bond (Agilent Poroshell C8-SB) and a mobile phase consisting of a gradient acetonitrile and formate buffer (3 mM of ammonium formate and 0.1% formic acid). The UPLC system was equipped with a triple quadrupole detector, which allowed the development of a "multiple reaction monitoring" (MRM) method for the analysis of TMAO. In detail, once thawed at room temperature and after centrifugation at 6000 rpm for 5 min, 25 μL of urine sample were diluted in 950 μL of UPLC mobile phase (1/1 (*v*/*v*) acetonitrile/ultra-pure sterile water + 0.025% of formic acid), and 25 μL of deuterated internal standard solution (1 ppm, TMAO-d9, Spectra 2000) were used for the normalization of results [26]. The UPLC samples were prepared mixing 950 μL of mobile phase [1/1 (*v*/*v*) acetonitrile/ultra-pure sterile water + 0.025% formic acid), 25 μL of urine sample, and 25 μL of deuterium-labeled methyl d9-TMAO solution (1 ppm; Spectra 2000 S.r.l., Roma, Italy). Mobile phase: 1/1 (*v*/*v*) acetonitrile/ultra-pure sterile water + 0.025% of formic acid. The run time per sample was 8 min. Sample freezing and thawing or their prolonged storage at room temperature did not have an impact on the TMAO quantification. A triple set of working standards of TMAO (trimethylamine N-oxide dihydrate, Fluka) at concentrations of 5 ppm, 50 ppm, 100 ppm was prepared according to the method described above, replacing the 25 μL of urine sample with 25 μL of standard solution. The average response factor was used for calculation.

#### *2.9. Statistical Analysis*

Statistical analyses of data were carried out using R statistic software (version 3.4.2). Concerning *cutC* gene and TMAO data, intrasubject variability was defined "high" when variance among the three replicates results were higher than twice the median of all variances. Correlation analyses were performed using the Kendall and Spearman formula with the items specified in the text as predictors and dependent variables. Significance was set at *p* ≤ 0.05, and mean differences in the range 0.05 < *p* < 0.10 were accepted as trends. To find associations among TMAO levels, bacterial taxa relative abundance, and *cutC* gene abundance, the machine learning supervised linear mixed model (LMM) algorithm was used. In brief, the LMM was performed using "lmer" function in the "lme4" library [27]. All samples were used in the LMM analysis (n = 48), considering that three measurements were available for each subject. The Akaike's Information Criterion (AIC) was used to test the goodness of fit of the LMM. The AIC index/value depends on the ANOVA test results between two models: The model that considered the effect of the predictors and the null model.

#### **3. Results**

#### *3.1. Distribution of the* cutC *Gene among Bacterial Taxa*

According to the literature, the ability of intestinal bacteria to convert the choline moiety to TMA is primarily associated with a recently discovered choline utilization (cut) genetic region harboring the *cutC* gene, which encodes a glycyl radical enzyme catalyzing C–N bond cleavage [15,18]. For this reason, we designed primers specifically targeting the *cutC* gene. These primers were intended for quantitative PCR (qPCR) experiments, and we avoided the use of degenerations in their sequence. In contrast, to target all putative *cutC* sequences identified in GenBank, we clustered the putative *cutC* genes into two groups (named Dd and Kp) according to sequence similarity (Supplementary Figure S1) and designed a pair of primers for each group in the most conserved sequence regions. Group Dd included putative *cutC* genes from *Firmicutes* (*Anaerococcus*, *Clostridium*, *Enterococcus*, *Streptococcus*), *Proteobacteria* (*Desulfotalea*, *Desulfovibrio*, *Enterobacter*), and *Actinobacteria* (*Olsenella*). Group Kp comprised putative *cutC* gene sequences from *Proteobacteria* (*Aeromonas*, *Enterobacter*, *Erwinia*, *Escherichia*, *Klebsiella*, *Pectobacterium*, *Pelobacter*, *Proteus*, *Providencia*, *Raoultella*, *Serratia*) and *Firmicutes* (*Desulfosporosinus*, *Enterococcus*).

Subsequently, the two primer sets were used in endpoint PCR reactions to test the presence of putative *cutC* genes within the genomic DNA isolated from the pure cultures of 64 bacterial strains. We obtained an amplicon of the expected size from seven strains. Specifically, strains *Streptococcus dysgalactiae* 485, *S. dysgalactiae* 486, and *S. dysgalactiae* A1.3 gave a band of the expected size with primers cut-Dd. In addition, strains *Enterococcus gilvus* MD179, *Enterococcus hirae* MD160, *Klebsiella oxytoca* MIMgr, and *Klebsiella* sp. MIMgr were positive with primers cut-Kp (Figure 1A,B). MS and NMR analyses revealed the ability to metabolize choline and produce TMA only for the same seven strains that resulted in positive PCR experiments (Figure 1C and Supplementary Figure S2).

**Figure 1.** Detection of the choline-utilization activity in pure bacterial cultures. Panels (**A**,**B**) represent agarose gel resulting from end-point PCR with primers cut-Dd (**A**) and cut-Kp (**B**). Panel (**C**) summarizes the detection of TMA in cell-free broth by mass spectrometry (MS) and nuclear magnetic resonance (NMR); +, TMA detected; -, TMA not detected. Lanes: **1**, *Escherichia coli* 3.1; **2**, *Lactococcus garvieae* FMBgr; **3**, *Enterococcus gilvus* MD160; **4**, *Enterococcus hirae* MD179; **5**, *Klebsiella oxytoca* MIMgr; **6**, *Klebsiella* sp. A1.2; **7**, *Streptococcus dysgalactiae* 485; **8**, *Streptococcus dysgalactiae* 486; **9**, *Streptococcus dysgalactiae* A 1.2; NC, negative control (i.e., M17 broth incubated without bacteria).

#### *3.2. Bacterial Taxonomic Structure of the Fecal Microbiota*

The metagenomic DNA isolated from the feces collected at three time points from 16 healthy adults (n = 48) was used in 16S rRNA gene profiling experiments. A total of 12,588,795 filtered high-quality sequence reads were generated with an average of 13,340 ± 8677 (mean ± standard deviation; max-min 11,594–4570) per sample.

We failed to stratify samples according to the 16S rRNA gene profiling data, indicating that fecal bacterial community structure was homogeneous among samples and among subjects (Supplementary Figure S3). In addition, we also observed that the overall composition of the fecal microbiota in each subject remained mostly stable over the three collection time points (Supplementary Figure S3). Globally, 182 bacterial genera were estimated, with a minimum of 36 and a maximum of 98 genera per fecal sample. *Bacteroides* was the most prevalent genus, followed by four genera of the order *Clostridiales* (undefined *Ruminococcaceae*, undefined *Lachnospiraceae*, *Ruminococcus*, and *Faecalibacterium*)

(Supplementary Figure S4A). At the family level, most of the reads were ascribed to only three families, i.e., *Ruminococcaceae*, *Bacteroidaceae*, and *Lachnospiraceae* (Supplementary Figure S4B).

#### *3.3. Putative* cutC *Genes in Human Fecal Metagenomic DNA*

In order to investigate the presence of *cutC* genes in the human gut microbiome, the cut-Dd and cut-Kp primer sets were used in qPCR experiments using the same fecal metagenomic DNA as a template from healthy adults used for microbiota profiling. All analyzed fecal samples gave a positive signal in qPCR with both primer pairs (Figure 2). In general, cut-Kp was detected at a higher relative concentration than cut-Dd (median ΔΔCt of 5.33 and 0.85 for cut-Kp and cut-Dd, respectively) (Figure 2A,B). In addition, with both cut-Kp and cut-Dd, six volunteers out of 16 showed a variance among the three replicates that was higher than twice the median of all variances, indicating a higher intrasubject variability (Figure 2A,B).

**Figure 2.** Fecal levels of the *cutC* gene and daily urinary excretion of trimethylamine-N-oxide (TMAO). The relative abundance of *cutC* was determined by quantitative real-time polymerase chain reaction (qPCR) with the primer pair cut-Dd-F/R (panel **A**) and cut-Kp-F/R (**B**). The TMAO concentration was determined by ultra-performance liquid chromatography coupled with tandem mass spectrometry (UPLC-MS/MS) in urine collected over 24 h (**C**). Green bars represent the mean ± standard deviation of three measurements per subject.

Subsequently, we performed correlation analyses between the *cutC* abundances determined with qPCR and the 16S rRNA gene profiling data to find potential relationships between the choline TMA-lyase genes and specific bacterial taxa of the fecal microbiota. To this end, we used the median relative abundance of bacterial taxa in fecal samples as predictors, whereas the dependent variables considered were the median abundances of cut-Dd and cut-Kp determined by qPCR per subject. We found that cut-Dd was positively correlated with taxa belonging to the phylum *Firmicutes*, including an undefined *Mogibacteriaceae* genus, *Oscillospira*, and the family *Christensenellaceae*. On the contrary, cut-Dd was negatively correlated with the *Firmicutes* order *Bacillales*, the *Firmicutes* genus *Streptococcus*, and the *Proteobacteria* genus *Haemophilus* (Supplementary Figure S5). Conversely, cut-Kp was positively associated with *Proteobacteria*. In particular, inside this phylum, a significant correlation was found with the family *Enterobacteriaceae* (Supplementary Figure S5).

#### *3.4. Daily Urinary Excretion of TMAO*

Subjects were asked to collect 24-h urine specimens the same days when the fecal samples were taken. Then, the levels of TMAO were quantified by UPLC-MS in all urine samples, revealing wide variability among the investigated healthy adults, with levels of urinary TMAO excretion ranging from less than 1 mg to more than 175 mg per day (Figure 2). We also observed an evident intrasubject variability in five volunteers whose TMAO excretion showed a variance among the three replicates that was higher than twice the median of all variances (Figure 2C). In particular, four out of the five volunteers with wide intrasubject variability (i.e., S07, S11, S19, and S22) were found to possess high intrasubject variability for *cutC* gene levels determined in qPCR experiments (Figure 2).

#### *3.5. Associations among Urinary TMAO, Fecal* cutC*, and Fecal Bacterial Taxa*

A linear mixed model was used to infer potential significant relationships among the datasets collected from volunteers at the three time points considered (Figure 3). TMAO was significantly associated with the cut-Kp/cut-Dd synergy (*p* < 0.001). Furthermore, studying the association of the single *cut* gene types, we observed that the relationship with TMAO was mainly determined by cut-Kp (Figure 3). In addition, we found a significant association between TMAO and 23 operational taxonomic units (OTUs). Conversely, cut-Kp and cut-Dd were significantly associated with 18 and eight OTUs, respectively. Notably, most of the OTUs that were significantly associated with cut-Kp (i.e., 15 out of 18) were also associated with TMAO, confirming the relationship between these two variables. Nine of the identified OTUs belonged to the phylum *Bacteroidetes*, while the remaining 21 were ascribed to *Firmicutes*. In addition, 80% of the OTUs (n = 24) belonged to only three families: *Bacteroidaceae*, *Lachnospiraceae*, and *Ruminococcaceae*. In particular, the most significant association (i.e., *p* < 0.001) referred to *Bacteroides caccae*, an undefined *Lachnospiraceae* genus, and several undefined *Ruminoccaceae* species (for TMAO and cut-Kp), *Bacteroides fragilis*, and an undefined *Clostridiales* species (for cut-Kp only) and an *Oscillospira* species (for cut-Dd) (Figure 3).

**Figure 3.** Analysis of the associations among fecal *cutC* gene abundances, fecal bacterial operational taxonomic units (OTUs), and urinary excreted TMAO carried out through a linear mixed model (LMM). Only OTUs that showed a significant association with *cutC* or TMAO are reported. The heatmap on the right represents TMAO levels, and *cutC* gene and OTU relative abundances. White boxes in the blue-yellow-red heatmap indicate that the OTU was not detected in that specific sample. The taxonomic lineage of each taxon is shown: p, phylum; c, class; o, order; f, family; g, genus; s, species. The black-yellow heatmap represents the Akaike's information criterion (AIC) values of the LMM analysis. Asterisks indicate significant associations: \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001; <sup>+</sup>, *p* < 0.1. syn. cut-Kp/Dd = synergy between cut-Kp and cut-Dd in LMM analysis.

#### **4. Discussion**

A growing number of studies have linked host TMAO levels to different diseases or prepathological metabolic states [28,29]. Conversely, TMAO has also been proposed as a beneficial factor that may promote protein stabilization and protect cells from osmotic and hydrostatic stresses according to a compensatory response mechanism [30]. The biological role of TMAO is therefore still debated. Nonetheless, a growing number of scientific studies have suggested that this molecule may play an important role in health and diseases [6].

It has been suggested that an important contribution to the hepatic production of TMAO is given by the TMA produced in the gut by microbial degradation of TMA-containing dietary molecules [31]. In particular, TMAO levels and their physiological consequences were shown to be significantly affected by the TMA derived from choline [15]. In this context, we studied the levels of TMAO excreted daily with urine, the composition of the intestinal microbiota, and the abundance of the choline TMA-lyase gene *cutC* in a group of healthy adult subjects with an Italian dietary pattern. The aim of this observational study was to verify whether TMAO levels excreted with the urine might be associated with the relative abundance of specific bacterial taxa and the bacterial gene *cutC* in feces. Literature focusing on the relationship among these three elements, particularly in non-diseased populations, is limited and partially contradicting [15,17–19].

The gene *cutC*, encoding the lyase enzyme essential for the conversion of choline into TMA [32], is not evenly distributed across bacterial taxa due to gene loss and horizontal gene transfer events that differently involve strains within the same species [15,18,33]. Therefore, predicting the choline degradation potential of a microbial ecosystem solely based on the taxonomic composition has many intrinsic limitations. The use of primers selectively targeting a specific enzymatic conserved domain may overcome this problem, permitting the selective quantification of the abundance of a gene coding for a specific enzymatic activity in the metagenomic DNA. A similar approach was used

by Martinez-Del Campo et al., who designed degenerate primers for the PCR amplification of the *cutC* gene from fecal metagenomic DNA and single strains [18]. The use of degenerate primers was necessitated by the fact that the CutC protein possesses sequence heterogeneity. In particular, Martinez-Del Campo et al. showed that the amino acid sequences deduced from the predicted bacterial *cutC* genes can be clustered into two groups (clades 1 and 2, [18]), which correspond to the CutC types I and II identified by Jameson et al. within a neighbor-joining phylogenetic tree constructed from amino acid sequences of glycyl radical enzymes [32]. The same result was found in our study by generating a distance tree based on the nucleotide sequences of putative *cutC* genes (Supplementary Figure S1). In particular, cluster cut-Dd corresponded to clade 1 and CutC type I, whereas cut-Kp included sequences coding for putative proteins found in clade 2 and CutC type II reported by the authors of [18] and [32], respectively.

For this reason, we developed two nondegenerate primer pairs located at the level of the catalytic site of the encoded enzyme that were useful for the amplification in (q)PCR experiments of the two clusters of the gene *cutC*.

When the two primer sets were used with the DNA of single strains, the only positive amplification signals were obtained with the bacteria that demonstrated the ability to metabolize choline in the biotransformation assay and produce TMA, confirming the suitability of these molecular probes to target choline-TMA-converting bacteria. Specifically, the bacterial strains identified here as able to degrade choline to TMA include species previously confirmed to exert this conversion, such as *Streptococcus dysgalactiae* [18]. In addition, we found *Klebsiella oxytoca*, which was reported to harbor a putative cut gene cluster [34], but has never been confirmed phenotypically. We also identified two positive *Enterococcus* strains. Reportedly, TMA production from choline has also been described for some enterococci, but not for the species *E. gilvus*, which is often isolated from food matrices, including meat, milk, and cheeses [35,36], and for the zoonotic pathogen *E. hirae* [37].

The qPCR experiments conducted showed that putative bacterial *cutC* genes were present in the fecal samples of all healthy adult subjects investigated. The high prevalence of this bacterial gene in the human gut microbiome was reported in a previous study, in which the presence of *cutC* homologs was observed in 96.6% of the assembled stool metagenomes of healthy individuals from the Human Microbiome Project (HMP) [18].

Reportedly, most of the TMA produced in the gut is absorbed into the portal circulation by passive diffusion [38]. Then, approximately 95% of the absorbed TMA is oxidized in the liver by flavin monooxygenases and excreted in the urine within 24 h [31,39]. Therefore, in this study, we performed a quantification of TMAO levels in urine samples obtained by 24-h collection.

The data presented here revealed a marked variability of both *cutC* and TMAO levels over the three time points considered in approximately 40% of volunteers. This instability was plausibly due to the variability of the daily food consumption of each subject. In this study, volunteers were free to follow their usual diet. Therefore, the analysis of multiple time points at approximately one-week intervals was useful to address the observed temporal instability of these parameters. To the best of our knowledge, this is the first work to report the stability of intestinal *cutC* and urinary TMAO levels over time.

This study has several limitations:


carnitine monooxygenase CntAB and the glycine betaine reductase GrdH, in addition to the choline TMA-lyase CutC [40].


However, despite the limitations described above, this study showed that changes in urine TMAO levels are associated with changes in the fecal abundance of the *cutC* gene and variations in the relative abundance of several bacterial taxonomic units of the fecal microbiota. In particular, TMAO was significantly associated with the levels of a specific subcategory of the *cutC* gene, which we named cut-Kp here. This result could be explained by the relative abundance of cut-Kp, which, by qPCR results, was approximately six-times higher than that of cut-Dd. According to correlation analysis, the most important contribution to cut-Kp gene abundance is provided by *Proteobacteria*, particularly by *Enterobacteriaceae*. This result is supported by the fact that cut-Kp has been quantified with primers designed on a cluster of gene sequences having the *cutC* of the *Enterobacteriaceae* species *K. pneumoniae* as a reference. Reportedly, the analysis of human gut metagenomes revealed a high proportion of the genera *Klebsiella* and *Escherichia*, which harbor three potential TMA-producing pathways, suggesting the importance of these bacteria for TMA cycling in the human gut [44].

Most OTUs that were found to be significantly associated with TMAO also had cut-Kp, confirming the relationship between TMAO and cut-Kp levels. A few OTUs were also associated with cut-Dd. All the taxonomic units associated with TMAO and *cutC* belong to only two taxonomic orders, *Bacteroidales* and *Clostridiales*. In particular, almost all the OTUs are attributable to only three families: *Bacteroidaceae*, *Lachnospiraceae*, and *Ruminococcaceae*. Notably, these families have been identified as the most metabolically active bacteria of the human microbiota and play a dominant role in the colonic fermentation of dietary fibers [45,46]. Reportedly, many of these bacteria do not display choline-utilization activities (e.g., cut genes have never been identified in *Bacteroidetes* and *Faecalibacterium*). Nonetheless, we can hypothesize an indirect association of these bacteria with *cutC* and TMAO based on the speculation that the higher presence of these bacteria might determine a greater utilization of the available nutritional sources in the colon, reducing substrates for the remaining bacterial communities. The latter may then receive selective pressure for the expansion of the activities related to the metabolization of the residual energy and carbon sources such as choline, resulting in increased TMA production.

#### **5. Conclusions**

Here, we described the results of a preliminary method-development study, which suggests the existence of a relationship between the levels of TMAO excreted in urine, some intestinal taxonomic groups belonging to the most active bacterial families of the colonic microbiota, and a subgroup of the *cutC* gene ascribable to the choline-TMA conversion enzymes of *Enterobacteriaceae*, named cut-Kp, whose relative abundance can be determined with the qPCR protocol developed in this study. Nonetheless, considering the limitations listed above, particularly concerning dietary intake, it is plausible to hypothesize that the results of this study may vary in other populations.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6643/12/1/62/s1, Figure S1: UPGMA hierarchical clustering based on ClustalW alignment of amino acid sequences of the choline trimethylamine lyase CutC. Figure S2. Verification of choline utilization and TMA production by single bacterial strains. Figure S3: Bacterial community structure of fecal samples. Figure S4: Tukey box and whisker plots representing the most abundant genera (A) and families (B) detected by 16S rRNA gene profiling in fecal samples collected from the adult volunteers participating in this study. Figure S5: Correlations among the fecal relative abundances of the choline TMA-lyase gene *cutC* and bacterial taxa. Table S1: Bacterial strains used for the screening of choline utilization activity. Table S2: Basic characteristics of the study participants.

**Author Contributions:** Conceptualization, S.G. and P.R.; methodology, S.G., A.D.V., V.T., I.V., V.G., G.R., E.R., C.G., V.D.V. and A.P.; formal analysis, S.G. and G.G; investigation, A.D.V., V.T., G.R., I.V., G.L.V. and A.P.; resources, S.G., V.G. and A.P.; data curation, S.G., A.D.V., V.T., G.G., and A.P.; writing—original draft preparation, S.G.; writing—review and editing, all authors; visualization, S.G., G.G. and A.D.V.; supervision, S.G, V.T. and P.R.; project administration, S.G.; funding acquisition, S.G., P.R. and V.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was partially funded by the University of Milan Funding "Linea 2-2016", PRISM Project. We acknowledge the European Joint Programming Initiative "A Healthy Diet for a Healthy Life" (JPI-HDHL—http://www.healthydietforhealthylife.eu/) and Mipaaft (Italy; D.M. 8245/7303/2016) for contributing to the grant awarded to G.G.

**Acknowledgments:** We thank Renata Piccinini for kindly providing the *S. dysgalactiae* strains.

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

#### **References**


© 2019 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*
