Altered gut microbiota due to daily intake of flavonoid-rich orange juice regulate depressive symptoms

Background : Depression is not just a general mental health problem but a serious medical illness that can worsen without treatment. The gut microbiome plays a major role in the two-way communication system between the intestines and brain. The current study examined the effects of flavonoids on depression by observing the changes in the gut microbiome and depressive symptoms of young participants consuming flavonoid-rich orange juice. Results : All participants (Depression) were evaluated using the Center for Epidemiological Studies Depression Scale (CES-D), a psychiatric screening tool used to detect preexisting psychiatric disorders, and were randomly divided into two groups: the orange juice (Depression_O) and flavored placebo (Depression_F) groups. For 8 weeks, flavonoid-rich orange juice (157.9±1.4 mg/100 g) or flavored placebo drink (28.4±0.7 mg/100 g) were consumed twice daily before breakfast and dinner. In total, 80 fecal samples from 40 participants (mean age, 21.83 years) were sequenced. Regarding depression, positive correlations were observed between brain-derived neurotrophic factor and the Lachnospiraceae family ( Lachnospiraceae_uc and Murimonas ) before flavonoid-rich orange juice treatment. Most notably, the abundance of the Lachnospiraceae family ( Lachnospiraceae_uc , Eubacterium_g4, Roseburia_uc, Coprococcus_g2_uc, Agathobacter_uc) had increased after flavonoid-rich orange juice treatment compared to that after flavored placebo treatment. We also validated the presence of unclassified Lachnospiraceae through sensitive real-time quantitative polymerase chain reaction using stool samples from participants before and after flavonoid treatment. Conclusions : Our results provide novel interventional evidence that alteration in the microbiome due to flavonoid treatment is related to potential improvement in depression in young adults. study the Clinical Research Service (CRIS)

may also result in suicide. It is being increasingly recognized that few clinical symptoms are found in healthy populations [2]. It is estimated that the number of individuals with depression worldwide is 350 million, and the population affected by depression is gradually expanding [1,3].
Dietary patterns and depression are closely related [4,5]. Previous studies have reported that the intake of sweets, red and processed meat, high-fat dairy products, refined grains and low intake of fruits and vegetables may increase the risk of depression [5][6][7]. Particularly, flavonoids and polyphenols (micronutrients) are abundant in fruits, vegetables, tea, and cocoa. Previous studies have revealed that flavonoid intake increases the amount of blood delivered to the brain, thereby improving brain cognitive function [8][9][10][11]. Epidemiological studies [12,13] and clinical studies [14,15] have also reported that high flavonoid intake through vegetables and fruits helps lower the risk of depression. Although fruits and fruit juices can easily supply flavonoids, such as hesperidin and narirutin, studies on their effectiveness on the human body are still at an early stage, and further studies are needed.
There have been increasing number of observational studies on the associations between habitual quality of diet and the prevalence of depression [16]. Additionally, it has been reported that dietary flavonoids (polyphenols) are associated with gut microbiome regulation, and these microbiomes can prevent and treat depression by increasing the synthesis of serotonin, which is a neurotransmitter in the body [16][17][18]. An experimental animal model showed that probiotics may reduce proinflammatory immune responses and increase levels of the serotonergic precursor, tryptophan, thereby ameliorating depressive symptoms [19]. Furthermore, depression can be reversed by administering the probiotic Bifidobacterium infantis to male Sprague-Dawley rats (n = 20) [20].
Tryptophan, as a precursor, is an essential amino acid needed by the body to synthesize serotonin [18]. As an essential amino acid, tryptophan is not synthesized from simple substances in humans and other animals; thus, it must be obtained through protein-based foods and dietary proteins.
It plays a major role in two-way communication between the intestines and brain [26,27]. Moreover, the impact of probiotics on human psychiatric disorders has recently emerged as an area of interest in neuroscience [28].
The activity of microbial metabolites is mainly due to their ability to permeate the blood-brain barrier.
Dietary polyphenols are metabolized by microorganism in the colon, and microbial metabolites of polyphenol, isoflavones, and lignans generally show greater permeability through the man-made gut and blood-brain barrier than their parent compound [29]. Another polyphenol metabolite, gallic acid derivatives, demonstrated neuroprotective effects through the modulation of the nuclear factor κB (NF-κB) pathway [30]. Polyphenols may affect the composition of the gut microbiota, and bioactive compounds in polyphenol metabolites produce clinical benefits [31].
Although previous studies have indicated that flavonoids consumed through the diet could affect gut microbiome regulation and depression, there have been limited intervention studies on the microbial metabolites of polyphenol. Therefore, this study aimed to evaluate the effects of flavonoids on depression in young adults by observing the changes in the gut microbiome and depressive symptoms of participants consuming flavonoid-rich orange juice. Furthermore, this study attempted to establish a theoretical basis for providing basic data and policy direction to develop a dietary program for alleviating and preventing depressive symptoms.

Characteristics of the study participants
The study participants (40) consisted of 16 men (40%) and 24 women (60%). The mean age was 21.83 ± 2.43 years, and there was no dropout. Statistical analysis was conducted for participant characteristics. Baseline characteristics did not differ between participants in the flavonoid-rich orange juice (Orange) group and low-flavonoid drink (Flavored) groups (Supplement Table 1). Table 1 Anthropometric data and blood test results at baseline and 8 weeks after intervention.

Variables
Orange (n = 20)  Table 2). The results of the 24-h recall showed that in the Orange group, only the intake of energy, protein, fat, phosphate, sodium, and vitamin B12 decreased significantly (p < 0.05), and that of folate and vitamin C increased significantly (p < 0.05). In the Flavored group, the intake of riboflavin and niacin significantly decreased (p < 0.05), while that of other nutrients did not change. The intake of folate and vitamin C intake was significantly (p < 0.05) different between the Orange and Flavored groups (Supplement Table 3).

Comparison of hematological profiles and anthropometric measurements
Blood test results revealed an increase in brain-derived neurotrophic factor (BDNF) levels in the Orange group (Table 1). The numerical increase in the levels in the Orange group was larger than that in the Flavored group but was not significantly different (p < 0.05). The serotonin level in the Orange group was high but was not significantly higher than that in the Flavored group, and there was no significant difference between the two groups (p = 0.058). Folate and homocysteine levels showed significant differences at baseline and post-intervention in the Orange group (p < 0.05). Homocysteine intake in the Flavored group was significantly different and fell out of the normal range (5.0-15.0 µmol/L). Moreover, after adjusting for sex, age, household income, and education level, body fat (%), homocysteine level, and high-sensitivity C-reactive protein (hs-CRP) level were significantly (p < 0.05) different between the Orange and Flavored groups ( Table 1).

Comparison of the Center for Epidemiological Studies Depression Scale scores
Post-intervention, the mean Center for Epidemiological Studies Depression Scale (CES-D) scores in the Orange and Flavored groups decreased to < 20 points (Table 1). Moreover, adjusted (sex, age, education, and household income) multiple regression analysis showed that the p-values of the CES-D score in both the Orange and Flavored groups decreased significantly after the intervention (p < 0.0001).
Sequencing characteristics and changes in microbial diversity in patients with depression In total, 80 fecal samples from 40 participants before and after the intervention were sequenced on the Illumina MiSeq sequencer. The sequencing results were obtained from 40 untreated participants (CES-D score ≥ 21), who were randomly divided into the Orange group (20 participants) and Flavored group (20 participants) for the analysis.
Phylogenetic alpha-diversity indexes (Chao, Phylogenetic Diversity, Shannon, and Simpson) were used to assess gut microbial diversity in participants with depression before and after flavonoid treatment while Faecalibacterium, Streptococcus, and Eubacterium_g23 abundance rates were lower in the Orange group ( Fig. 2A). Further, we also used metagenome analysis to validate the linear discriminant analysis (LDA) effect size (LEfSe) and found that eight families (Bifidobacteriaceae, Akkermansiaceae, Bacteroidaceae, Veillonellaceae, Coriobacteriaceae, Lachnospiraceae, Lactobacillaceae, and Rikenellaceae) were abundant in the Orange group ( Fig. 2A). An LDA analysis (p < 0.05, LDA score > 2) showed that ten taxons were more abundant in the Orange group (Fig. 2B).
Lachnospiraceae (Roseburia, Coprococcus_g2, Agathobacter, Clostridium_g24, and Eubacterium_g4), Bifidobacteriaceae (Bifidobacterium), and Bacteroidaceae (Bacteroides), which were enriched in the Orange group, were the major phylotypes that contributed to the difference in the gut microbiota composition between the Depression_O and Orange groups.

Changes in microbiota taxonomic composition in the Depression_F and Flavored groups
The gut microbiota composition of 20 individuals with depression (before flavored treatment group [Depression_F]) and 20 participants in the Flavored groups showed differences in two phyla, Firmicutes and Bacteroidetes. The relative abundance of Firmicutes was greater in the Flavored group than in the Depression_F group, while Bacteroidetes had a lower abundance rate in this group ( Fig. 3A). At the family level, Ruminococcaceae, Bacteroidaceae, and Bifidobacteriaceae were less abundant in the Flavored group, but Lachnospiraceae, Coriobacteriaceae, Streptococcaceae, and Lactobacillaceae were more abundant in this group. Bacteroides and Bifidobacterium were enriched in the Depression_F group compared with that in the Flavored group at the genus level (Fig. 3A). The LEfSe analysis also showed that the Bacteroidetes phylum and Bacteroides and Subdoligranulum genera were enriched in the Depression_F group, while the Firmicutes phylum and Blautia, Streptococcus, and Coprococcus genera were abundant in the Flavored group (Fig. 3A). LDA analysis (p < 0.05, LDA score > 2) revealed that Lachnospiraceae (Blautia_uc and Bifidobacterium_uc) were more abundant in the Flavored group than in the Depression_F group (Fig. 3B).

Association between gut microbiota and depression
We also evaluated the associations of serum biomarkers with the relative abundances of gut microbiota. We found that depression and serum biomarkers were closely associated with gut microbiota in the Depression group (Fig. 4). The serum BDNF level was significantly positively correlated with the abundance of the Lachnospiraceae family (Lachnospiraceae_uc and Murimonas) and was negatively correlated with OCTT_g (Ruminococcaceae) abundance in those with depression.
The relative abundances of Fusicatenibacter, PAC001043_g, and Eubacterium_g20 were positively correlated with body mass index (BMI), while those of the Ruminococcaceae family (PAC001100_g and OCTT_g) and Christensenella genus were negatively correlated with BMI (p < 0.001). Additionally, homocysteine levels were positively correlated with the abundance of the Gemella genus (p < 0.01) in patients with depression.
Flavonoid-rich drink increases the relative abundance of Lachnospiraceae_uc and

Bifidobacterium_uc in patients with depression
To explore the contribution of the flavonoid-rich orange drink to the observed changes in gut microbial compositions in those with depression, we conducted a real-time polymerase chain reaction (PCR) experiment with stool samples from the Depression group and Orange and Flavored groups. We used the universal bacterial gene for copy control and another normal stool sample (cohort) for internal control ( Table 2). The results of real-time PCR showed that the relative abundance of Lachnospiraceae_uc and Bifidobacterium_uc was significantly increased in the Orange group compared to that in the Depression_O group ( Fig. 5A and C). However, the relative expression of Lachnospiraceae_uc was increased and that of Bifidobacterium_uc was decreased in the Flavored group compared with those in the Depression_F group, but there was no significant difference ( Fig. 5D and F). The relative expression of Roseburia_uc in the Orange group was increased compared with that in the Flavored group, but no significant difference was noted ( Fig. 5B and E).

Discussion
This intervention study characterized the gut microbiota in the Depression (CES-D scores ≥ 21, untreated) group compared with those in the flavonoid-rich versus placebo treatment groups.   Fig. 1A and 2A). Moreover, significantly increased expression of Lachnospiraceae_uc and Roseburia_uc were observed in the Orange group compared with that in the Flavored group as shown by real-time PCR results of stool samples (Fig. 5). Thus, we found that Lachnospiraceae responds to flavonoid treatment in patients with depression.
In the Orange group, the relative taxonomic abundance of Bifidobacterium, Roseburia, Ruminococcus, and Akkermansia significantly increased compared to that in the Flavored group (Supplement Figs. 1 and 2). The relative expression of Bifidobacterium_uc in the Orange group was also significantly increased compared to that in the Flavored group ( Fig. 5C and F elative abundances of bacterial genera, including Akkermansia spp., were significantly reduced in socially defeated animals, which was positively correlated with both anxiety and depression [45].
Moreover, the administration of prebiotics increased the relative abundance of Akkermansia during exposure to stressors [46,47]. The abundance of Akkermansia has been suggested as a biomarker for healthy intestines and has an inverse correlation with several intestinal disorders [48]. Generally, Akkermansia and Bifidobacterium are well-known health-associated genera that protect against inflammation, promote immunomodulation, and promote healthy metabolic homeostasis [49]. These findings corroborate our study on flavonoid intervention, and the relative abundance of Akkermansia significantly increased in the Orange group compared to that in the Flavored group.
Greater microbial diversity was found in the Depression group than that in the Orange and Flavored In our study, the abundances of Ruminococcaceae and Lachnospiraceae were positively correlated with BDNF levels in the Depression group. After the 8-week intervention, the serum levels of BDNF in the Orange group were higher than those in the Flavored group. The abundances of Ruminococcaceae OCTT_g (p < 0.0016), Ruminococcus_g2 (p < 0.007), Ruminicoccus (p < 0.0248), and Lachnospiraceae_uc (p < 0.039) were positively correlated with serum BDNF levels. BDNF is a neurotrophin that performs multiple functions in the central nervous system and participates in the therapeutic mechanisms of antidepressants [53]. BDNF levels were significantly lower in patients with MDD than the control group, and recovery from depression after antidepressant treatment was associated with normal serum levels of BDNF in patients with MDD [54]. Similarly, there were more changes in the taxonomic composition of the microbiota in the Orange group than in the Flavored group. In the Orange group, 10 taxons, including Lachnospiraceae, Bacteroides, and Bifidobacterium, were more abundant, but only one taxon (Blautia_uc) increased in abundance in the Flavored group compared to that in the Depression_F group. Besides, the taxonomic abundance of Bifidobacterium was decreased in the Flavored group ( Fig. 3F and Supplement Fig. 2B). Moreover, the relative abundance of Clostridium decreased further in the Orange group (Supplement Fig. 1F and 2F). The prevalence of Clostridium was negatively associated with serum BDNF levels [38]. Therefore, we found that the 8-week intervention changed the gut microbiota composition and increased serum BDNF levels in the Orange group compared to those in the Flavored group.
Members of the Lachnospiraceae family, such as the Roseburia, Blautia, and Coprococcus genera, are known to break down carbohydrates into short-chain fatty acids (SCFAs) [55]. Reduction in these fermentation-related bacteria precipitates a decline in SCFA production, causing intestinal barrier dysfunction [55,56]. Moreover, SCFAs promote the differentiation of T cells and can function as a histone diacetylase inhibitor [57]; therefore, SCFAs can act as a regulator of immune homeostasis.
Additionally, SCFAs are involved in neurotransmitter production [58] and neuroprotection and can penetrate the blood-brain barrier [59]; therefore, they have been proposed as potential novel antidepressants [60]. The Ruminococcaceae family was highly abundant in healthy controls compared to that in patients with MDD [38]. Moreover, the Ruminococcaceae family is characterized by antiinflammatory activity and associated with a chronic low-grade inflammatory response [49,61].
Therefore, the relative abundance of these genera in the gut microbiome mediated the low degree of inflammation and higher intestinal barrier function. Inflammation is associated with major depression [62]. Furthermore, flavonoids exhibit a neuroprotective effect by falsifying inflammatory reactions and have potential therapeutic effects in terms of neuroprotection [63].
In this study, the use of antibiotics, probiotics, and prebiotics was not allowed for assessing the microbial community. Additionally, we served fresh, 100%-pure, orange juice daily as the flavonoidrich drink and investigated the associations between the relative abundance of the gut microbiome and depression in young adults. Although the particular association between flavonoids and depression is still unclear, we found that high intake of flavonoids changed the relative abundance of the gut microbiome and increased serum BDNF levels. An unidentified genus of Lachnospiraceae, Bacteroides, and Bifidobacterium was correlated with flavonoid-rich orange juice treatment.

Conclusion
we demonstrated that the intake of flavonoid-rich orange juice for 8 weeks has an influence on depression in young adults. Moreover, the change in gut microbiota composition due to flavonoid-rich orange juice treatment was associated with the butyrate-producing Lachnospiraceae family.
Therefore, we suggest the efficacy of flavonoid-rich orange juice treatment in improving depression.

Methods Participants
All procedures were conducted in accordance with the Declaration of Helsinki (1996)  Daily orange juice intake was monitored by phone and SMS, one to one by researchers. There was no dropout during the study.
All participants provided consent for participation after understanding the objective of this study, test schedule, and potential risk. Those who were using antibiotics, had a family history of mental illness, had been diagnosed with psychiatric disorders, had a history of antipsychotic medication use, or had a bowel disease were excluded from the study.

Intervention study design
The study had a randomized, single-blind design and was conducted for 8 weeks. The volunteers were randomly divided into two groups of 20 participants each, the Orange and Flavored groups, Beverage Co. Ltd., Korea) as the LF drink (190 mL). The total flavonoid content was analyzed by highperformance liquid chromatography mass spectrometry. The total flavonoid content in the orange drink was 157.9 ± 1.4 mg/100 g and that in the flavored drink was 28.4 ± 0.7 mg/100 g (Supplement Table 4). The two drinks were almost identical in appearance, volume, taste, calories, and glucose, fructose, and sucrose levels, except in the flavonoid content (Supplement Table 5).

Blood tests
Participants were asked during screening to avoid certain foods (e.g., high-fat and high-sucrose foods, berries, fruits, fruit juice, tea, jams, and alcohol) and to have an LF diet for 24 h before the blood test.
Fasting blood samples (5 mL of blood) were collected at the beginning (baseline) and at the end (8 weeks) of the treatment. Serum obtained from each blood sample was divided into 500 µL aliquots and was used to measure BDNF, serotonin, folate, homocysteine, hs-CRP, and vitamin B12 levels.

Dietary intake
During the study period, the FFQ and 24-h dietary recalls were assessed twice by a qualified dietician.
Using FFQ, the frequency and amount of food consumed by each participant during the past year were examined.
No separate major dietary intervention or physical activity program was offered. Moreover, participants were advised to maintain physical activity for the duration of the study.

Bioinformatics analysis
Raw reads were checked for quality, and low-quality reads (< Q25) were filtered using Trimmomatic 0.32 [52]. After the quality control process, paired-end sequence data were merged together using PANDAseq [69]. Then, primers were trimmed using a proprietary program of ChunLab at a similarity cutoff of 0.8. Nonspecific amplicons that did not encode 16S rRNA were detected by HMMER's hmmsearch program with 16S rRNA profiles [70]. Sequences were denoised using DUDE-Seq, and non-redundant reads were extracted by UCLUST clustering [71,72]. The EzBioCloud database was used for taxonomic assignment using USEARCH (8.1.1861_i86linux32) followed by more precise pairwise alignment [72,73]. UCHIME [74] and the nonchimeric 16S rRNA database from EzBioCloud were used to detect chimeras on reads with a best hit similarity rate < 97%. Sequence data were then clustered using CD-HIT [75] and UCLUST [72]. The alpha-and beta-diversity analyses with the Chao,

Associations of the gut microbiota with BDNF levels and clinical indexes in patients with depression
We assessed the associations of biomarkers and BMI with the gut microbiota in patients with depression. The correlation coefficient (r) between gut microbiota and biomarkers was calculated using Spearman's rank correlation analysis. The heat map was plotted in 35 genera for participants with BDNF levels and statistically significant results (p < 0.1). All statistical analyses were performed using the SAS 9.4.

Statistical analysis
The Student's t-test, paired t-test, and Wilcoxon signed-rank test were performed using SPSS (version 23.0; SPSS Inc., USA). Results are expressed as mean ± Standard Error of the Mean (SEM), and p values less than 0.05 were considered to be statistically significant. p values of less than 0.05, 0.01, and 0.001 are described as * p < 0.05, ** p < 0.01, and *** p < 0.001, respectively.

Consent for publication
This work was supported by the National Research Foundation of Korea (NRF) and a grant funded by the Korean government (MSIT) (No. NRF-2016R1D1A1B03935435).

Author Contributions
Investigation and data arrangement, M.P. and J.C. Writing, and original draft preparation, M.P. and J.C.    Associations of the gut microbiota with BDNF levels and clinical parameters in patients with depression. The heat map of Spearman's rank correlation coefficients obtained by comparing clinical parameters (BDNF scores, p < 0.05) and relative abundances of gut microbiota in patients with depression. The heat map was plotted using R software. +, p < 0.10; ++, p < 0.05; +++, p < 0.01.