Microbial Metabolites of Flavan-3-Ols and Their Biological Activity

Flavan-3-ols are the main contributors to polyphenol intake. Many varying beneficial health effects in humans have been attributed to them, including the prevention of cardiovascular disease and cancer. Nevertheless, the mechanisms by which these flavonoids could exert beneficial functions are not entirely known. Several in vitro studies and in vivo animal models have tried to elucidate the role of the specific colonic metabolites on the health properties that are attributed to the parent compounds since a larger number of ingested flavan-3-ols reach the colon and undergo there microbial metabolism. Many new studies about this topic have been performed over the last few years and, to the best of our knowledge, no scientific literature review regarding the bioactivity of all identified microbial metabolites of flavan-3-ols has been recently published. Therefore, the aim of this review is to present the current status of knowledge on the potential health benefits of flavan-3-ol microbial metabolites in humans while using the latest evidence on their biological activity.


Introduction
Polyphenols exert numerous beneficial health effects in the human organism. The main contributors to polyphenol intake are flavan-3-ols [1,2], which constitute the most complex flavonoid subclass. After ingestion, flavan-3-ols can reach the colon in high proportions, where they can be transformed into microbial metabolites by the resident microbiota. These catabolites might be responsible for the health promoting effects that are attributed to the parent compounds ( Figure 1). Monagas et al. laboriously complied all of the available scientific information related to the metabolic pathway of flavan-3-ols in the human organism and reviewed the biological activities that were attributed to their phase-II and microbial metabolites [3]. However, since then, many new studies regarding this topic have been performed and, to the best of our knowledge, no other scientific literature review about the bioactivity of all of the identified microbial metabolites of flavan-3-ols to date has been recently published. Only Mena et al. recently addressed the bioactivity of phenyl-γ-valerolactones and phenylvaleric acids: two of the several groups of flavan-3-ol colonic metabolites [4]. Thus, the aim of this review is to summarize the most recent scientific publications on biological activity of microbial metabolites of flavan-3-ols to present the current status of knowledge on this topic.

Origin
Flavan-3-ols are present in food not only in the form of monomers, but also as oligomeric and polymeric proanthocyanidins, whose degree of polymerization can reach up to 50 units or more [5].

Microbial metabolites
Phenyl-γ-valerolactones and phenylvaleric acids Low molecular weight phenolics Monagas et al. laboriously complied all of the available scientific information related to the metabolic pathway of flavan-3-ols in the human organism and reviewed the biological activities that were attributed to their phase-II and microbial metabolites [3]. However, since then, many new studies regarding this topic have been performed and, to the best of our knowledge, no other scientific literature review about the bioactivity of all of the identified microbial metabolites of flavan-3-ols to date has been recently published. Only Mena et al. recently addressed the bioactivity of phenyl-γ-valerolactones and phenylvaleric acids: two of the several groups of flavan-3-ol colonic metabolites [4]. Thus, the aim of this review is to summarize the most recent scientific publications on biological activity of microbial metabolites of flavan-3-ols to present the current status of knowledge on this topic.

Origin
Flavan-3-ols are present in food not only in the form of monomers, but also as oligomeric and polymeric proanthocyanidins, whose degree of polymerization can reach up to 50 units or more [5].
They are commonly found in foods, such as fruits, green tea, red wine, and chocolate [6], although the main dietary source in the population differs demographically and, therefore, the type of flavan-3-ol that is consumed also differs [7]. Their original content in food can also vary due to several factors, such as the plant variety, the time of harvest, environmental factors, processing, and storage conditions [6]. For this reason, flavan-3-ol content in foods has been detected at very different concentrations ranging, for example, between 6 and 544 mg/100 g green tea, or between 77 and 273 mg/100 g dark chocolate [7].
These flavonoids have been shown to exert beneficial health effects in humans in observational as well as in intervention studies. For example, their intake might prevent cardiovascular diseases [8] or cancer [9]. Nevertheless, the mechanisms by which they could have this positive impact on health are not entirely known. In fact, the bioavailability of flavan-3-ols might be lower than 4 % in humans [10].
Several factors could have an influence on bioavailability, such as food matrix, compound concentration, or gut microbiota composition. Additionally, the degree of polymerization influences their bioavailability: generally, monomeric flavan-3-ols can be absorbed in the small intestine, while the structures with a higher polymerization degree are metabolized by the intestinal microbiota before they are finally absorbed [3].
Therefore, the fact that a large proportion of flavan-3-ols reach the colon suggests that they might not be the active substances that exert beneficial physiological effects.

Biological Properties of Flavan-3-Ol Microbial Metabolites
After the unabsorbed flavan-3-ols reach the colon, free monomeric forms are released and then become available for microbial transformation (Figure 1). These free monomers are converted by specific bacteria into diphenylpropan-2-ols before being further transformed to phenyl-γ-valerolactones, which are intermediate metabolites that are exclusive to flavan-3-ols ( Figure 2). Further reactions, which include breaking the valerolactone ring and dihydroxylation, form various forms of phenylvaleric acids [3,4]. After absorption, colonic metabolites undergo phase II metabolism in the liver and their conjugated derivatives reach the organs and tissues, where they will exert their beneficial effects. They can also be eliminated in the urine and faeces [3].
Unlike the native monomeric and, especially, the oligomeric and polymeric flavan-3-ols, colonic metabolites are highly bioavailable in the human organism. In fact, their absorption has been shown to occur at a much higher rate than in the case of the parent compounds [10,[13][14][15]. Phenyl-γ-valerolactones The further bacterial metabolism of phenyl-γ-valerolactones and phenylvaleric acids produces different hydroxylated forms of phenyl and benzoic acids by the loss of carbon atoms from the side chain through ß-oxidation ( Figure 2) [3]. The 3,4-dihydroxylated phenolic acids are dehydroxylated at C-4' and C-3', producing 3-and 4-monohydroxylated phenolic acids [3]. Metabolites, like vanillic acid, homovanillic acid, hippuric acid, or p-coumaric acid, have also been related to the intestinal catabolism of flavan-3-ols ( Figure 2) [3]. These low molecular weight phenolic acids are, in contrast to phenyl-γ-valerolactones, not specific to flavan-3-ols, since they can also be formed after the metabolism of other flavonoids [11,12].
After absorption, colonic metabolites undergo phase II metabolism in the liver and their conjugated derivatives reach the organs and tissues, where they will exert their beneficial effects. They can also be eliminated in the urine and faeces [3].
Unlike the native monomeric and, especially, the oligomeric and polymeric flavan-3-ols, colonic metabolites are highly bioavailable in the human organism. In fact, their absorption has been shown to occur at a much higher rate than in the case of the parent compounds [10,[13][14][15]. Phenyl-γ-valerolactones and phenylvaleric acids have been previously found at concentrations lower than 1 µM in plasma [15][16][17][18][19]. The smaller phenolics have been mainly found at concentrations lower than 0.5 µmol/L [16][17][18][19], but in some cases, such as phenylacetic acid, protocatechuic acid, and hippuric acid, the concentrations could reach ≈ 40 µmol/L [16,20]. Table 1 presents an overview of the studies that were performed over the last few years regarding the biological activities, and including the physiological importance, of phenyl-γ-valerolactones and phenylvaleric acids that, according to the available literature, are related to the human colonic metabolism of monomeric flavan-3-ols and proanthocyanidins [3,5].

Cardiovascular Protective Effect
Phenyl-γ-valerolactones and phenylvaleric acids could contribute to the preventive effects that were attributed to flavan-3-ols on cardiovascular diseases through hypotensive properties and by attenuating the monocyte adhesion to endothelial cells that are involved in the development of atherosclerosis.
Particularly, two metabolites of (−)-epigallocatechin gallate produced by intestinal microbiota, namely 5-(3,4,5-trihydroxyphenyl)-γ-valerolactone and 5-(3,5-dihydroxyphenyl)-γ-valerolactone, were reported to have hypotensive effects in vivo in rats [23]. The first metabolite that was administered at a concentration of 150 mg/kg significantly decreased systolic blood pressure 2 h after intake when compared to the control group [23]. In contrast, the second metabolite required a higher oral dosage of 200 mg/kg to exert an effect on systolic blood pressure 4 h after its administration [23].
Recently, Lee et al. reported that monocyte-endothelial cell adhesion was dose-dependently prevented after treatment with 5-(3',4'-dihydroxyphenyl)-γ-valerolactone at concentrations under 30 µM [24]. Two of the main molecules that were involved in the monocyte-endothelial adhesion are vascular cell adhesion molecule (VCAM)-1 and monocyte chemoattractant protein (MCP)-1, and their tumor necrosis factor (TNF-α)-stimulated expression was downregulated after treatments with the phenyl-γ-valerolactone at concentrations between 7.5 and 30 µM [24]. This was possibly due to the parallel downregulation of the phosphorylation of two nuclear factor kappa B (NF-κB) activation signalling regulators as well as NF-κB transcriptional activation [24].

Immunostimulatory Activity
Some of the positive effects of the intermediate microbial metabolites from flavan-3-ols seem to be as well mediated by direct stimulation of the immune system. In this regard, Kim et al. reported that 5-(3',5'-dihydroxyphenyl)-γ-valerolactone plays an important role on immunostimulation, because the splenic NK cell cytotoxic activity against murine lymphoma YAC-1 cells was significantly increased after its oral administration (10 mg/kg) in comparison to the control and (−)-epigallocatechin treatment groups [27]. In addition, 5-(3',5'-dihydroxyphenyl)-γ-valerolactone upregulated the splenic production of IFN-γ, a T cell growth factor and effector of CD4 + T cells [27]. CD4 + T cell activity also increased after incubation with 10 µM 5-(3',5'-dihydroxyphenyl)-γ-valerolactone for 72 h, as well as it did with other microbial metabolites lacking the 4'-hydroxyphenyl group on the B ring [27].

Neuroprotective Effect
The in vitro stimulation of nerve cell proliferation and differentiation by phenyl-γ-valerolactones was recently reported by Unno et al. [28]. Particularly, only 0.05 µM 5-(3',5'-dihydroxyphenyl)-γ-valerolactone could enhance human SH-SY5Y neuroblastoma cell number, but at concentrations that were higher or equal to 1.0 µM, the effect was reduced [28]. The same concentration of the microbial metabolite, including its sulfated form, increased the length and number of neurites in SH-SY5Y as compared to control cells, whereas the glucuronide form could only increase the number of neurites [28]. Table 2 presents the biological activities reported for smaller phenolic acids result of microbial catabolism of flavan-3-ols.  [32]

Phenolic Acids
Antidiabetic effects Glucose transport in human and murine 3T3-L1 adipocytes stimulated or not with insulin PCA 100 µmol/L PCA reversed the oxLDL-induced drop in glucose uptake and GLUT4 translocation. PCA also prevented the oxLDL-induced reduction of adiponectin mRNA expression and secretion, as well of PPARγ mRNA expression and activity. [33] Beta cell function of rat INS-1E pancreatic beta cells and isolated rat pancreatic islets 3,4-DHPA; 2,3-DHB and 3-HPP 1-5 µM 3,4-DHPA and 3-HPP significantly increased glucose-induced insulin secretion (5 and 1 µM, respectively). In presence of oxidative stress, 3,4-DHPA and 3-HPP reduced ROS and carbonyl group production, and glucose-stimulated insulin secretion was restored to control levels. The phosphorylation of PKC and ERKs was enhanced. [34] Beta cell function of Min6 pancreatic beta cells incubated with cholesterol 3,4-DHPA 10-250 µM 3,4-DHPA prevented impaired insulin secretion induced by cholesterol by protecting pancreatic beta cells against oxidative stress, apoptosis and mitochondrial dysfunction. [35] Insulin signalling and glucose uptake and production in rat renal NRK-52E cells 2,3-DHB; 3,4-DHPA; 3-HPP and VA 20 µM Glucose uptake and production decreased after treatment with 2,3-DHB, and PEPCK levels as well. IR and IRS-1 phosphorylated and total protein levels were increased. The inhibition of the PI3K/Akt pathway was restrained. [36] Insulin signalling and glucose uptake and production in rat renal NRK-52E cells treated with high glucose 3,4-DHPA; 2,3-DHB and 3-HPP 10 µM 3,4-DHPA restored the altered glucose uptake and production caused by high glucose, and tyrosine phosphorylated and total levels of IR increased. The PI3K/Akt pathway and AMPK were activated, while the PEPCK expression was decreased. [37] Beta cell function and glucose utilization in human skeletal muscle and rat INS-1 beta cells HA; HVA and 5-PVA 5-100 µM HA and 5-PVA stimulated glucose oxidation in skeletal muscle and preserved skeletal mitochondrial function after oxidative insult.
In beta cells, all metabolites induced glucose-stimulated insulin secretion without affecting beta cell mitochondrial respiration or electron transport chain components' expression.

PCA 25 µM
OxLDL-induced cell death was prevented, as well as ROS production and GSH depletion. The activation of p53 was prevented, and therefore the overexpression of p53-target genes decreased. p38MAPK and PKC∂ activation was reversed. PCA induced JNK activation and increased nuclear Nrf2 content. [51] TGF-ß1-induced proliferation and migration of human airway smooth muscle cells (ASMCc) PCA 1-50 nM PCA inhibited the proliferation and migration of ASMCs and the expression of type I collagen and fibronectin. The Smad2/3 activation in ASMCs exposed to TGF-ß1 was downregulated. [52] Cardiovascular protective effect Antithrombotic efficacy under high shear stress in vitro in human platelets as well as in an in vivo arterial thrombosis model PCA

5-25 µM
PCA significantly decreased stress-induced platelet aggregation by blocking the interaction between von Willebrand factor (vWF) and glycoprotein Ib. Intracellular calcium increase was attenuated, shear-induced granular secretion from dense and α-granules was inhibited and glycoprotein IIb/IIIa activation was attenuated.
The antithrombotic effects of PCA were confirmed in vivo. [53] Cardiac function and cardiac autonomic balance in STZ-induced diabetic rats PCA 50 and 100 mg/kg %FS and %LVEF increased and LF:HF decreased compared with untreated diabetic rats. Plasma HbA1c decreased as well as cardiac MDA and cardiac mitochondrial ROS. Mitochon-drial membrane depolarization and swelling was prevented and cardiac anti-apoptotic BCL2 protein levels increased [54] Vasodilation of pre-contracted isolated aortic rings; blood pressure in normotensive and spontaneously hypertensive rats  Relaxation of pre-contracted rat artery rings and blood pressure in spontaneously hypertensive rats [62] Angiogenesis in HUVEC cells treated with VEGF and in zebrafish model PCA inhibited the proliferation, migration, invasion and capillary structure formation of HUVECs, and blocked the VEGFR2-dependent Akt/MMP2 and ERK pathways. In vivo, the anti-angiogenic effect of PCA was possibly due to downregulation of VEGFα-VEGFR2 and Ang2-Tie2 pathways.  Concentration-dependent enhancement of the ALDH activity, as well as expression of ALDH1A1, ALDH2 and ALDH3A. Nuclear levels of Nrf2 and AhR increased significantly at 20 µM, while those of NF-κB decreased.
[68] Menadione-induced liver damage in rats PCA 10 and 20 mg/kg PCA prevented menadione mediated-alterations in hepatocellular markers and it also increased the activities of antioxidant enzymes (SOD and CAT) and phase II detoxifying enzymes (GST and NQO-1), and Nrf-2. [69]

Modulation of intestinal microbiota
Alteration of the composition in fecal microbiota of ApoE-/-mice fed on a high-fat diet FA 30 mg/kg FA significantly lowered the ratio of Firmicutes to Bacteroidetes when compared to obese control group. [72] Lipid metabolism in HFD-induced obesity in mice FA 30 mg/kg Serum TC, TG and LDL-C decreased when compared to obese control group, and liver TC and TG levels as well. FA significantly increased the mRNA expression of AHR and decreased that of FAS and SREBP-1c.    LPS-mediated bone loss in vivo was also restored by PCA. [79] Osteoclast differentiation and apoptosis in RAW264.7 murine macrophage cells treated with RANKL PCA 8 µM PCA inhibited osteoclast differentiation by regulating oxidative stress and inflammation, and induced apoptosis in mature osteoclasts by inducing mitochondrial membrane potential, Cyt c release and caspase activation. [80] Renoprotective effects Redox status in high-glucose-exposed rat renal proximal tubular NRK-52E cell

Antidiabetic Effect
The antidiabetic effects of various microbial metabolites from flavan-3-ols have been investigated from different perspectives, and the involved mechanisms are apparently not only related to their antioxidant properties, but also to their ability to modulate different signalling pathways.
Scazzocchio et al. assessed the role of protocatechuic acid in the glucose transport in adipocytes [33]. Incubation of the metabolite at 100 µM for 18 h with human and murine adipocytes treated with oxidized LDL (oxLDL) showed that protocatechuic acid significantly reversed the detrimental effects of oxLDL on glucose uptake and glucose transporter type 4 (GLUT4) translocation. In addition, when both adipocytes were insulin-stimulated and also when human adipocytes were not insulin-stimulated, protocatechuric acid significantly removed the detrimental effect of oxLDL on adiponectin secretion [33]. Glucose uptake increased in a significant and concentration-dependent manner in non-oxLDL-treated human and murine adipocytes in the absence of insulin up to 40% and 60%, respectively [33]. These results suggested an insuline-like activity. The reduced peroxisome proliferator-activated receptor-γ (PPARγ) mRNA expression and activity induced by oxLDL was also counteracted, and its inhibition blocked both the adiponectin and GLUT4 upregulation, which suggests a direct involvement on the insuline-like activity of protocatechuic acid [33].
The effect of low molecular weight phenolics on beta cell functionality has also been assessed.

Fernández-Millán et al. reported a significant increase in glucose-induced insulin secretion in INS-1E
pancreatic beta cells and isolated rat islets after treatment with 3,4-DHPA and 3-hydroxyphenylpropionic acid (3-HPP) at low concentrations (5 and 1 µM, respectively) [34]. Moreover, in the presence of oxidative stress that was induced by t-BOOH, both 3,4-DHPA and 3-HPP restored glucose-stimulated insulin secretion to control levels and significantly reduced t-BOOH-induced cell death as well as ROS and carbonyl group production [34]. Protein kinases, specifically protein kinase C (PKC) and extracellular signal-regulated kinases (ERK), could be involved, since their phosphorylation levels were enhanced after treatment and their pharmacological inhibition blocked the increased insulin secretion that was induced by the metabolites [34].
Carrasco-Pozo et al. evaluated the protective effect of 3,4-DHPA on Min6 pancreatic beta cells dysfunction that was induced by high cholesterol, and the metabolite at 250 µM could indeed prevent the decrease in insulin secretion induced by high cholesterol [35]. Furthermore, it prevented the cholesterol-induced cytotoxicity and apoptosis in a concentration-dependent manner, and also prevent oxidative stress and mitochondrial dysfunction [35]. Nuclear factor (erythroid-derived 2)-like 2 (Nrf2) translocation to the nucleus was increased by around six-fold relative to control, which suggests that it contributes to the antioxidant protection exerted by 3,4-DHPA [35].
Recently, Bitner et al. reported that hippuric acid and homovanillic acid, as well as the phenylvaleric acid 5-phenylvaleric acid, more effectively enhanced glucose-stimulated insulin secretion (GSIS) in beta cells than (−)-epicatechin at concentrations between 5 and 100 µM (hippuric acid between 5 and 50 µM) [38]. Contrary to (−)-epicatechin, the microbial metabolites enhanced GSIS without enhancing beta cell mitochondrial respiration or increasing expression of mitochondrial electron transport chain components [38]. In addition to the stimulation of beta cell function, the microbial metabolites stimulated glucose utilization in skeletal muscle and they preserved mitochondrial function after insulation [38].
Álvarez Cilleros et al. investigated how flavan-3-ol-derived low molecular weight phenolics influenced the mechanisms related to the glucose homeostasis in rat renal NRK-52E cells since kidneys are involved in the maintenance of glucose homeostasis [36]. The authors found out that 2,3-dihydroxybenzoic acid (2,3-DHB) at 20 µM reduced cellular glucose uptake in a similar way to the sodium-glucose co-transporter-2 (SGLT-2) antagonist phlorizin, without altering the expression levels of SGLT-2 and glucose transporter-2 (GLUT-2) [36]. Glucose production was significantly reduced and the levels of the gluconeogenic enzyme phosphoenolpyruvate carboxykinase (PEPCK) were also reduced [36]. In addition to the effects on glucose uptake and production, the effect of the microbial metabolite on the activation of proteins from the insulin signalling pathway was evaluated. 2,3-DHB significantly increased insulin receptor (IR), insulin receptor substrate 1 (IRS-1) tyrosine phosphorylated, and total protein levels, as well as the phosphorylated levels of protein kinase B (Akt) and glycogen synthase kinase-3 (GSK3), and subsequently reduced the glycogen synthase (GS) phosphorylated levels [36]. Moreover, the authors showed that Akt was involved in the modulation by 2,3-DHB of both PEPCK levels and glucose production in the NRK-52E cells [36].
Similarly, in another study from Álvarez-Cilleros et al., when renal tubular NRK-52E cells were treated with high glucose levels, 3,4-DHPA at 10 µM restored the induced alteration in glucose uptake and production and increased tyrosine phosphorylated and total levels of IR [37]. Moreover, it restrained the inhibition of the phosphatidylinositol-3-kinase (PI3K)/Akt pathway that was involved in the insulin signalling cascade and counteracted the high glucose-induced downregulation on 5 -AMP-activated protein kinase (AMPK) phosphorylation and upregulation on PEPCK [37]. When Akt and AMPK were inhibited the protective effect of 3,4-DHPA was abrogated, which suggests that they play a key role on the preservation of renal tubular functionality, the modulation of the glucose homeostasis by the metabolites, together with the attenuation of the insulin signalling blockade [37].

Antiglycative Activity
Advanced glycation end-products (AGEs) contribute to the occurrence of several pathologies, such as diabetes and rheumatoid arthritis [83]. For this reason, the effect of phenolics against their formation has been evaluated. Dihydroferulic acid has been shown to significantly inhibit albumin glycation at a concentration of 10 µmol/L, and the combination of 3-hydroxyphenylacetic acid (3-HPA), 3,4-DHPA, and homovanillic acid at 2.0 µmol/L has also shown this [39]. 3,4-DHPA at 1 mM exerted an inhibitory activity against AGEs that was significantly lower than that of rutin and quercetin, but significantly higher than the positive control aminoquanidine, 3-HPA and homovanillic acid [40].

Anti-inflammatory Activity
Prevention of inflammation by microbial low molecular weight phenolics has been shown to occur by modulating the mediators of inflammation and additionally the signal transduction pathways involved.
The secretion of another inflammatory mediator, TNF-α, induced by LPS in THP-1 monocytes, was reduced by benzoic acid-sulfate, vanillic acid-glucuronide, and protocatechuic acid-3-sulfate in a study from di Gesso et al. [42]. Moreover, four different combinations of metabolites that included 4-hydroxybenzoic acid or protocatechuic acid or both could also significantly reduce TNF-α secretion, but, interestingly, to a greater extent than the metabolites alone, which showed no significant effect [42]. Wang et al. showed that protocatechuic acid at 5, 10, and 20 µM dose-dependently inhibited the production of the inflammatory mediators TNF-α, interleukin 6 (IL-6), interleukin 1 beta (IL-1ß), and prostaglandin E 2 (PGE 2 ) in LPS-stimulated BV2 microglia [43]. The LPS-induced expression of toll-like receptor 4 (TLR4), activation of NF-κB, and mitogen-activated protein kinases (MAPKs) in BV2 microglial cells was furthermore suppressed by protocatechuic acid, which suggests that these signaling pathways are actively involved in its anti-inflammatory activity [43]. The same metabolite at 5, 10, and 20 µM was also shown to inhibit LPS-induced production of IL-6 and interleukin 8 (IL-8) in human gingival fibroblasts, secondarily to its ability to activate PPAR-γ [44].
In a 2,4,6-trinitrobenzenesulfonic acid (TNBS) model of rodent inflammatory bowel disease, protocatechuic acid administered at 30 and 60 mg/kg improved the TNBS-induced colitis and reduced oxidative stress that was measured as GSSG/GSH ratio [45]. The antioxidant enzymes expression increased after treatment with protocatechuic acid when compared to the TNBS group and additionally the transcription factor Nrf2 protein levels [45]. The TNBS-induced increase in colonic expression of the inflammatory markers IL-6, TNF-α, IL-1β, and cyclooxygenase-2 (COX-2) was reversed [45]. The SphK/S1P signaling pathway seems to play a key role, since TNBS-induced an increase in mRNA levels, an increase in protein concentration and immunohistochemical labelling for SphK1 was prevented by protocatechuic acid [45]. Similarly, the increase in S1P production and expression of S1P receptor 1 and S1P phosphatase 2 were significantly reversed [45]. Related signaling pathways might be also involved, since protocatechuic acid reversed the TNBS-induced increase of phosphorylated signal transducer and the activator of transcription 3 (STAT3), NF-κB p65 subunit expression, and phosphorylation of Akt and ERK [45]. In a similar model, gallic acid that was given at 20-60 mg/kg improved the TNBS-induced colitis in mice and inhibited inflammation reducing proinflammatory cytokines, such as IL-1 and IL-6, and increased the anti-inflammatory cytokines via inhibiting the NF-κB pathway [47].
Although the evaluation of the antioxidant activity of all these compounds has been performed while using different assays, as well as at different concentrations and conditions, it is clear that low molecular weight phenolics can act as antioxidants. As suggested by many authors, the number of hydroxyl groups can be involved in their antioxidant activity.

Anti-proliferative Activity and Cytotoxicity
The anti-proliferative activity of low molecular weight phenolics has been assessed in a variety of cellular models, including murine macrophages as well as human cancer cell lines. For a better classification of the biological activities of metabolites, the studies assessing the anti-proliferative activity on human cancer cell lines are presented in a following section about chemopreventive effects.
In a study from Varì et al., protocatechuic acid at a concentration of 25 µM counteracted the cytostatic and cytotoxic effect of oxLDL in J774A.1 macrophages [51]. The oxLDL induced an increased apoptosis as well as oxidative stress, and both were reduced after treatment with the metabolite, although the ROS production could only be abolished during the first six hours of treatment [51]. The metabolite at 25 µM could reduce the oxLDL-induced activation of p53 protein and counteract the overexpression of the main p53 target genes: p66Shc and Bax [51]. The activation of the c-Jun N-terminal kinases (JNK)/Nrf2 pathway could play a key role in the anti-apoptotic effects of protocatechuic acid, since its inhibition caused a change in the expression of phosphorylated p53, Bax and the active form of caspase-3 comparable to levels that were found in cells only treated with oxLDL [51]. This inhibition also suppressed the effect of protocatechuic acid on the reduction of oxLDL-induced ROS levels [51].
Liu et al. also recently reported the anti-proliferative effects of protocatechuic acid in preventing or treating asthma airway remodeling, since it suppressed the proliferation of airway smooth muscle cells and extracellular matrix protein deposition in transforming growth factor-beta1 (TGF-ß1)-mediated airway smooth muscle cells via the inactivation of Smad2/3 signaling pathway [52].

Cardiovascular Protective Effect
Low molecular weight phenolics could have a positive effect on the prevention of cardiovascular disease through protection of the heart and the endothelial function, through antithrombotic effects, as well as through their previously presented anti-inflammatory activity.
On the one hand, the anti-platelet and antithrombotic potential of protocatechuic acid were confirmed in a study from Kim et al. [53]. This metabolite at concentrations between 5 and 25 µM significantly decreased stress-induced platelet aggregation in isolated human platelets, which was mediated by blocking the interaction between von Willebrand factor (vWF) and platelet receptor glycoprotein Ib [53]. The increase of intracellular calcium induced by shear stress was dose-dependently attenuated by protocatechuic acid. Moreover, this metabolite inhibited shear-induced granular secretion from dense and α-granules and attenuated GP IIb/IIIa activation. Interestingly, protocatechuic acid did not inhibit platelet aggregation induced by other endogenous agonists [53]. In an in vivo rat arterial thrombosis model, the antithrombotic effects of PCA were confirmed and this metabolite did not show increased risk of bleeding [53].
Similarly, 4-methylcatechol has been reported to also have anti-platelet effects [60]. When testing its effect on whole blood platelet aggregation induced by arachidonic acid, this metabolite was much more active than acetylsalicylic acid (IC 50 = 3 µM vs. IC 50 = 25 µM, respectively) [60]. This could be confirmed by the authors in an ex ovo thrombotic model that mimics the in vivo situation [60]. Moreover, 4-methylcatechol was shown to interfere with calcium intracellular trafficking, and this seemed to be its main mechanism of action. As the authors suggest, the methyl group probably does not play any role in the anti-platelet potential, but the hydroxyl groups in positions 1 and 2 seem to be important [60].
On the other hand, the cardioprotective effects have been attributed to protocatechuic acid by Semaming et al. [54]. When administered to type 1 diabetic rats at a dosage of 50 and 100 mg/kg, it significantly increased the fractional shortening and the left ventricular ejection fraction when compared to non-treated diabetic rats after 8 and 12 weeks, and the effects were partially comparable to those after treatment with insulin [54]. Moreover, the low-frequency:high-frequency ratio was significantly decreased when compared with the non-treated diabetic rats. The authors also found a significant decrease of plasma HbA1c and cardiac MDA levels, an improvement of cardiac mitochondrial function, and an increase in anti-apoptotic BCL2 expression [54].
In another study, the same metabolite was shown to have a positive effect on diabetic cardiomyopathy in type 2 diabetic rats when orally administered (50 and 100 mg/kg) [56]. Protocatechuic acid exerted hypoglycemic and insulin-sensitizing activities via the stimulation of IRS1/PI3K/AKT/AMPK/GLUT4/P 38 signaling pathway in the skeletal muscle and anti-inflammatory effects mediated by the downregulation of the poly (ADP-ribose) polymerase (PARP)/PKC/NF-κB signaling cascade in the myocardial tissue [56]. Moreover, antioxidant and radical scavenging effects in the myocardial tissue of type 2 diabetic rats were also attributed to the administration of protocatechuic acid [56].
3-HPP was also reported to be beneficial for endothelial function by Qian et al. [57]. Particularly, 1 µM 3-HPP was able to maintain the NO production stimulated by insulin in human aortic endothelial cells under glucotoxic conditions, and additionally to increase endothelial nitric oxide synthase (eNOS) and Akt phosphorylation [57].
Álvarez Cilleros et al. also reported an increased NO production too, but, in this case after the treatment of EA.hy927 human endothelial cells with 10 µM 3,4-DHPA or with a mix of that metabolite with 2,3-dihydroxybenzoic acid (2,3-DHB) and 3-HPP (12 µM) [58]. AMPK and Akt seem to be key mediators, since their inhibition blocked the NO production as well as eNOS phosphorylation [58]. Under oxidative stress that is induced by t-BOOH, both 3,4-DHPA and the mix of the metabolites reversed the induced endothelial dysfunction by preventing increased ROS generation and the activation of signalling pathways related to oxidative stress [58].

Chemopreventive Effect
The potential cancer preventing effect of low molecular weight phenolics has often been studied in in vitro models of intestinal cell lines. Henning et al. found that the proliferation of HCT116 colon cancer cells was inhibited by 3,4-DHPA and 3-O-methylgallic acid [61]. The concentrations of 3,4-DHPA that were needed for a significant inhibition of cell proliferation were 200 and 300 µmol/L, but the concentration exhibiting 50% inhibition was higher than 400 µmol/L. However, 3-O-methylgallic acid, had a IC 50 = 260 µmol/L [61]. When 3,4-DHPA was combined with epigallocatechin, the inhibitory effects significantly increased when compared to the individual treatments [61].
Rosa et al. also reported 3,4-DHPA to induce the largest reduction of cell viability of HT-29 human colon adenocarcinoma cells at concentrations between 2.5 and 100 µM (up to 66% reduction) [50]. p-Coumaric acid, vanillic acid, and ferulic acid could also reduce cell viability [50]. The metabolites modulated the cell cycle, since 3,4-DHPA, p-coumaric acid, vanillic acid (10 and 100 µM), and additionally ferulic acid (10 µM) decreased the cell number in S phase [50]. Relative apoptosis rate data showed that p-coumaric acid and ferulic acid at 10.0 µM promoted increased apoptosis [50]. In a study from López de las Hazas et al., late-stage apoptosis of HT-29 colon cancer cells was reported to be also induced by 4-hydroxyphenylacetic acid [62].
Since angiogenesis, the process by which new blood vessels are formed is involved in the progression of tumors, Hu et al. evaluated the anti-angiogenic activity of protocatechuic acid in vitro as well as in an in vivo animal model [63]. In the in vitro model with human umbilical vein endothelial cells (HUVEC), protocatechuic acid inhibited the vascular endothelial growth factor (VEGF)-induced proliferation, the formation of capillary tubes, and the migration and invasion of HUVECs [63]. The mechanisms that are involved might be related to its ability to attenuate ROS production (6.25-100 µM) as well as by interfering with VEGF-dependent Akt/MMP2 and ERK pathways [63]. The in vivo model in zebrafish showed that, at 25 µM, protocatechuic acid also had a significant anti-angiogenic effect when compared with the control group, possibly through downregulation of the angiogenesis-related signal transduction pathways of VEGFα-VEGF Receptor 2 and angiotensin 2-Tie2 receptor [63].
Protocatechuic acid at up to 30 µM was furthermore able to reduce cell viability of ovarian cancer cells OVCAR-3 in a dose-dependent manner (IC 50 = 10.7 µM) and it induced apoptosis [65]. Furthermore, it induced the cell cycle arrest in G 2 /M phase [65]. The increased apoptosis was accompanied by the activation of PARP and caspase-3, the upregulation of Bax and downregulation of B-cell lymphoma 2 (Bcl-2) in OVCAR-3 cells [65]. Additionally, autophagy-related protein LC3-II was upregulated and the intracellular ROS levels decreased, while those of glutathione increased [65].

Modulation of Drug Metabolizing Enzymes
The correct functioning of drug metabolism is essential in transforming and removing foreign substances from the organism. The expression and functions of the enzymes taking part in this process can be altered in some physiological situations [84], and the enhancement of their capacity could protect against pathologies, such as cancer [85]. Therefore, the effect of phenolic compounds on drug metabolizing enzymes has been previously evaluated.
3,4-DHPA also modulated the gene expression of drug metabolizing enzymes as well in Hepa1c1c7 cells in a study from Tang et al. [49]. The mRNA levels of glutamate-cysteine ligase catalytic subunit (GCLC), heme oxygenase 1 (HO-1), and NADH: quinone oxidoreductase-1 (NQO1) significantly increased at concentrations of the metabolite of 5, 10, and 25 µM, respectively [49]. The mRNA levels of cystine/glutamate anti-porter (xCT) and Cytochrome P450 1A1 (CYP1A1) were additionally increased, but at higher concentrations of 100 and 250 µM, respectively [49]. The GSH-activity in RL34 normal hepatocytes was also increased in a dose-dependent manner, and reached an activity of 1.4-fold higher than control levels [49]. Moreover, 3,4-DHPA reduced the cytotoxic effect of peroxide in Hepa1c1c7 cells in a concentration-dependent manner, and cell viability was completely restored to control levels at a concentration of 100 µM [49].
Furthermore, Xue et al. reported that 3,4-DHPA was able to inhibit oxidative stress in acetaminophen-induced hepatotoxicity in rats by decreasing malondialdehyde levels and by dose-dependently increasing the antioxidant enzymes glutathione (GSH), glutathione peroxidase (GPx), glutathione S-transferase (GST), and superoxide dismutase (SOD) to prevent the decrease of UDP-glucuronosyltransferase (UGT) and sulfotransferase (SULT)'s activity induced by acetaminophen [67].
The effect of 3,4-DHPA on aldehyde dehydrogenase was also investigated by Liu et al., and the authors found that, after treatment of mouse hepatoma Hepa1c1c7 cells, the total activity of ALDH increased, as well as the expression of ADLH1A1, ADLH2 and ALDH3A1 [68]. The authors also showed that 3,4-DHPA activates the signalling pathways of Nrf2 and AhR to possibly increase the expression of the ALDH genes, but inhibits that of NF-κB [68].
In addition to 3,4-DHPA, protocatechuic acid was also shown to modulate drug metabolizing enzymes. Ibitoye et al. administered protocatechuic acid to rats at 10 and 20 mg/kg and it significantly reversed menadione-induced increases in superoxide ions and hydrogen peroxide in addition to decreases in the activities of SOD and catalase (CAT) [69]. The metabolite reversed the induced decrease in GST and NQO-1, and significantly increased Nrf-2, Akt, and PI3K, which suggests that this is a possible pathway for the increasing antioxidant and phase II metabolizing activities [69].

Modulation of Intestinal Microbiota
Ma et al. investigated the effect of one microbial metabolite, ferulic acid, on faecal microbiota of HFD-induced obese mice [70]. The authors reported that FA significantly reduced the ratio of Firmicutes to Bacteroidetes when compared to the obese control group [70].
The growth of two probiotic (Lactobacillus rhamnosus and L. acidophilus) and two pathogenic bacteria (Escherichia coli and Salmonella enterica serovar Typhimurium) in the presence of gallic acid, vanillic acid, ferulic acid, and protocatechuic acid was evaluated in another study [71]. The results showed that the minimal inhibitory concentration (MIC) and minimal bactericidal concentration (MBC) of phenolic acids against E. coli and S. Typhimurium showed similar values of 15-30 mmol/L [71]. However, the MICs were higher for probiotics (mostly >35 mmol/L), except in the case of ferulic acid (MIC = 20 mmol/L) [71]. The MBC were, in this case, higher than 35 mmol/L for all metabolites. Furthermore, the presence of the metabolites in MRS broth without dextrose promoted the growth of lactobacilli [71]. Therefore, the compounds selectively inhibited the growth of pathogenic bacteria without affecting the viability of probiotics [71].

Modulation of Lipid Metabolism
The effect of ferulic acid on lipid metabolism in high-fat diet-induced obese mice was studied by Naowaboot et al. [72]. The results show that this metabolite at a dosage of 25 and 50 mg/kg reduced the serum lipid level (total cholesterol, triglycerides, and non-esterified fatty acids), as well as liver cholesterol and triglyceride when compared to the obese mice control group [72]. This could be explained by the modulation of the lipogenic genes: the expression of sterol regulatory element-binding protein 1c (SREBP1c), fatty acid synthase (FAS), and acetyl-CoA carboxylase (ACC) were reduced by the metabolite [72]. Moreover, hepatic carnitine palmitoyltransferase 1a (CPT1a) gene and peroxisome proliferator-activated receptor alpha (PPARα) proteins were up-regulated, being involved in the stimulation of ß-oxidation genes [72].
Similarly, after a 12-week treatment of high-fat diet-induced obese mice with 30 mg/kg of ferulic acid, serum total cholesterol, triglycerides, and LDL cholesterol significantly decreased when compared to obese control group [70]. Total cholesterol and triglyceride levels in liver were as well significantly reduced after treatment with the metabolite [70]. The mechanism could be due to the significant increase of aryl hydrocarbon receptor (AHR)'s mRNA expression, which is thought to regulate lipid metabolism and inhibit FAS and SREBP1c. Indeed, after treatment with ferulic acid the mRNA expression of FAS and SREBP-1c decreased [70].
In a model of neuroinflammation, SH-SY5Y human neuroblastoma cells were exposed to SIN-1-induced nitrosative stress and treated with 3,4-DHPA, 3-HPP, and 3-HPA [74]. All metabolites at concentrations between 0.1 -10 µM enhanced cell viability when compared to control group, but 3,4-DHPA had the strongest effect at 10 µM not only after short time exposure, but also after more prolonged time exposure to SIN-1 [74]. The increase in caspase-3 activation induced by SIN-1 was also reduced when the SH-SY5Y cells were treated with 10 µM 3-HPP [74]. Possible mechanisms are the reduction of SIN-1-induced increase of p38 phosphorylation by 3-HPA and of ERK1/2 phosphorylation by all metabolites [74].
The same cell line was exposed to H 2 O 2 in a study from González-Sarrías et al. and treated with 3,4-DHPP, 3,4-DHPA, and gallic acid, among other polyphenol-derived metabolites [75]. The negative effects of H 2 O 2 on SH-SY5Y neuroblastoma cell viability were reverted and the percentage of late apoptosis decreased up to approximately 12% with 10 µM pretreatment or cotreatment with the metabolites as compared with H 2 O 2 treatment alone [75].
In primary cultures of cerebellar granule neurons, 4-hydroxybenzoic acid (100 and 200 µM) and protocatechuic acid (50-200 µM) protected cells from oxidative stress induced by H 2 O 2 [76]. Under conditions of nitrosative stress typical from inflammative situations in the central nervous system, only protocatechuic acid (10-300 µM) was neuroprotective [76]. However, under conditions of excitotoxicity only 4-hydroxybenzoic acid prevented neuronal death [76]. The ability of these metabolites to alter NO production in a BV2 microglial cell line following treatment with LPS was also tested, and only protocatechuic acid at 100 µM exerted anti-inflammatory activity reducing NO production [76].
In a lysolecithin (LPC)-induced model of inflammation in hippocampal neurons, gallic acid and valeric acid, at 1 and 0.2 µM, respectively, inhibited the LPC-induced demyelination and promoted the formation of myelin, as well as neurite outgrowth [78].
In vivo, the progress of hippocampal neurodegeneration in the brain of diabetic rats was reduced after the administration of 20 and 40 mg/kg gallic acid and p-coumaric acid, respectively [73]. These metabolites exerted antioxidant, anti-inflammatory, and anti-apoptotic activities, which probably mediated the protective effects on the hippocampus [73].
Kho et al. recently also evaluated in vivo the effects of protocatechuic acid at a dose of 30 mg/kg on global cerebral ischemia-induced hippocampal neuronal death in rats [77]. The protocatechuic acid-treated rats showed 51%, 58%, 75%, and 76% fewer degenerating neurons than in the vehicle-treated group in the hippocampal areas cornu ammonis 1, subiculum, cornu ammonis 3, and dentate gyrus, respectively [77]. Additionally, ischemia-induced blood-brain barrier disruption was prevented and the induced inflammatory responses mediated by microglia and astrocytes. Oxidative injury was reduced by the metabolite, and the GSH levels after the treatment increased 37% [77].

Osteoprotective Effect
Some low molecular weight phenolics could be effective in preventing bone resorption. In particular, protocatechuic acid was able to inhibit osteoclastogenesis and bone loss in in vivo and in vitro models in a study from Park et al. [79]. At concentrations between 1 and 25 µM, protocatechuic acid dose-dependently inhibited RANKL-induced osteoclast differentiation in mouse bone marrow macrophages and blocked the bone-resorbing activity of mature osteoclasts, probably by suppressing JNK signaling, c-Fos stability, and the expression of osteoclastic marker genes [79]. Mice that were treated with protocatechuic acid efficiently recovered from LPS-induced bone loss in vivo [79].
In another study from Wu et al., protocatechuic acid was also shown to suppress osteoclast differentiation and induce apoptosis in mature osteoclasts [80]. It dose-dependently reduced the Tartrate-resistance acid phosphatase (TRAP) activity induced by RANKL and multinucleated osteoclast formation at a concentration up to 10 µM, as well as ROS and lipid peroxide levels with an increase in antioxidant status [80]. Osteoclast specific markers and transcription factors' expression was significantly downregulated when compared to that of RANKL, as well as MAPK activation and the expression of inflammatory proteins [80]. Oxidative stress and inflammation were shown to be regulated by protocatechuic acid through regulating the transcription factor Nrf-2 [80]. Moreover, in mature osteoclasts it induced apoptosis by the loss of mitochondrial membrane potential, cytochrome c release, and caspase activation [80].

Renoprotective Effects
Several studies tried to define the preventive properties of microbial phenolic metabolites on diabetic nephropathy, which is a common complication in diabetic patients. In a recent study from Álvarez-Cilleros et al., the renal proximal tubular NRK-52E cells were treated under high glucose levels with 3,4-DHPA in order to assess its protective effect on the redox status [81]. 3,4-DHPA at 10 µM reversed the glucose-induced ROS production and the decrease of antioxidant defenses [81]. SIRT-1 and NADPH-oxidase-4 (NOX-4) appear to play a key role in 3,4-DHPA-mediated renal protection [81].
Protocatechuic acid could also have beneficial effects on renal disease, as shown by Ma et al. [82]. At concentrations between 5 and 10 µM, it dose-dependently inhibited the high glucose-induced proliferation of human mesangial cells and suppressed extracellular matrix expression (type IV collagen, laminin, and fibronectin) [82]. Similarly, the levels of ROS and malondialdehyde induced by high glucose were decreased [82]. The mechanistic proposed by the authors is related to the p38 mitogen-activated protein kinase (MAPK) signaling pathway, since protocatechuic acid downregulated their phosphorylation levels in high glucose-stimulated mesangial cells [82].

Predictive Potential
In addition to the fact that they are highly bioavailable, when studying the biological activities of flavan-3-ol microbial catabolites, it is important to consider their physiological concentration in plasma after their absorption. Many of the bioactivity studies performed used metabolite concentrations that were much higher than the ones reported in plasma [15][16][17][18][19][20], therefore making the results difficult to relate to health effects in humans.
However, it is worth mentioning that microbial low molecular weight phenolics, in contrast to phenyl-γ-valerolactones, can be formed not only from flavan-3-ols, but also after colonic degradation of many other polyphenols present in food. Therefore, they can be formed from different sources and absorbed in the colon in a continuous manner, which could suggest a higher plasma concentration in the actual physiological conditions. Interestingly, Uhlenhut and Högger reported a high binding capacity of a phenyl-γ-valerolactone to macrophages, monocytes, and endothelial cell, which made the authors suggest this intracellular accumulation to be the reason why in vivo bioactivity appears at plasma concentrations that exert no effect in vitro [22]. Some authors assessed the effects of mixtures of metabolites [26,39,42,58,61,62,64,71], and reported that some of these combinations had greater activities than the individual compounds. This is also important to consider a more realistic approach, since a huge variety of microbial metabolites are formed in the colon from flavan-3-ols and from other polyphenols, and are therefore present in the organism together and not isolated.
Furthermore, the bioactivity of not only the microbial metabolites, but also of their conjugated forms has been examined in several studies [21,28,42]. This is of unquestionable relevance, since the microbial catabolites formed in the colon can be further conjugated, and these forms might also contribute to the biological actions of the flavan-3-ols too.
Although the flavan-3-ol colonic metabolic pathway has been almost fully elucidated and many of their microbial metabolites have been identified, their formation has been shown to differ among individuals, as reviewed elsewhere [4]. These interindividual differences in the formation of microbial metabolites can be explained by differences in gut microbiota composition, and this is commonly termed in the literature as "metabotype". Since several biological properties have been associated with flavan-3-ol microbial metabolites, specific metabotypes producing these catabolites could be predictors of their health effects, as previously shown for other polyphenols [86][87][88].
While metabotypes for determined microbial metabolites from polyphenols, such as ellagitannins [89] and isoflavones [90], have been well studied and described, less is known about the gut metabolic phenotypes that are involved in the production of flavan-3-ol colonic metabolites.
Three metabotypes in the formation of flavan-3-ol microbial metabolites were recently proposed for the first time in a study by Mena et al. (n = 11). These were defined by the different production amounts of trihydroxyphenyl-γ-valerolactones, dihydroxyphenyl-γ-valerolactones, and hydroxyphenylpropionic acid [91]. In contrast, another study showed that there were no specific proanthocyanidin-derived microbial metabolites that enabled the stratification of individuals by metabotypes, which suggests no influence of interindividual gut microbiota differences on the diverging health effects of proanthocyanidins [92]. Due to the high number of subjects needed for the identification of metabotypes and the difficulty of using phenyl-γ-valerolactones as determinant for metabotypes [93], more research is needed in this area.

Conclusions
There are numerous available studies that suggest positive effects of flavan-3-ol microbial metabolites. However, caution needs to be taken when interpreting their results and extrapolating them to humans, since only cellular and animal models have been performed. In this regard, human in vivo studies of microbial metabolites may not be realistic either, since their oral administration by human beings would not reproduce their actual formation through microbial metabolism in the colon. Moreover, other factors need to be considered when evaluating the positive health effects of microbial metabolites, such as individual differences in their colonic production, bioavailability, and concentration in biological fluids and tissues.
Although the interest in microbial metabolites of flavan-3-ols has been growing over the last few years and many studies have been performed with the aim of understanding their biological effects in organisms, further research that clarifies their metabotype-dependent formation, their conjugation, and their pharmacokinetics patterns are needed in an attempt to harmonize the study methodologies and, ultimately, to elucidate their health benefits in human beings.