Citrus polyphenols and risk of type 2 diabetes: Evidence from mechanistic studies

Abstract

Citrus fruits are a rich source of (poly)phenols, a group of dietary bioactive compounds that protect against developing type 2 diabetes. Our review critically evaluates how experimental in vitro and animal models have elucidated some of the underlying mechanisms on how citrus (poly)phenols affect the markers of type 2 diabetes. According to animal studies, the beneficial effects derived from consuming citrus compounds appear to be related to long-term effects, rather than acute. There are some notable effects from citrus (poly)phenol metabolites on post-absorptive processes, such as modulation of hepatic glucose metabolism and insulin sensitivity in target tissues, but with a more modest effect on digestion and sugar absorption within the gut. Experimental studies on cells and other systems in vitro have indicated some of the possible mechanisms involved, but ∼70% of the studies utilized unrealistically high concentrations and forms of the compounds, compromising physiological relevance. Future studies should discuss the relevance of concentration used in in vitro experiments, relative to the proposed site of action, and also examine the role of catabolites produced by the gut microbiota. Finally, it is important to examine the relationship between the gut microbiota and bioavailability on the action of citrus (poly)phenols.

Keywords:

• Citrus fruits
• (poly)phenols
• starch digestion
• glucose metabolism
• in vitro
• animal studies

Introduction

According to the World Health Organization (WHO 2016), the global prevalence of diabetes has nearly doubled since 1980, rising from 4.7% to 8.5% in the adult population, of which 90–95% of the cases correspond to type 2 diabetes. High postprandial blood glucose fluctuations and insulin resistance are the key factors increasing the risk of type 2 diabetes (Bhupathiraju et al. 2014; Greenwood et al. 2013; Livesey et al. 2008, 2019). (Poly)phenols, a group of compounds naturally present in plants, attenuate some of the risk factors of type 2 diabetes, and also exhibit some anti-inflammatory effects (Abbas et al. 2017; Alam et al. 2014; Cao et al. 2019; Kalt et al. 2020; Kwon et al. 2007; Tripoli et al. 2007; Williamson 2017; Williamson, Kay, and Crozier 2018). Several epidemiological and human intervention studies suggest increased consumption of (poly)phenol-rich foods is associated with reduced risk of type 2 diabetes and cardiovascular disease (Domínguez-Avila et al. 2017; Gee and Johnson 2001; Shahidi and Peng 2018).

Citrus fruits are widely grown and consumed worldwide, and are a major source of vitamin C and various phytonutrients. Different in vitro, in vivo (animal), and human intervention studies have reported the possible role of citrus fruits and their (poly)phenolic components in attenuating metabolic syndrome (Cerletti et al. 2015; Constantin et al. 2014; Dong et al. 2016; Fujioka et al. 2006; Hwang et al. 2015; Jung et al. 2006; R. W. Li et al. 2006; O’Neil et al. 2012; Onda et al. 2013; de Paiva et al. 2019; Ponce, Benassi, and Cesar 2019). Among the citrus (poly)phenols, the flavanones, hesperidin 1, narirutin 5, and naringin 6 (Ahmed et al. 2017; Constantin et al. 2014; Jung et al. 2004, 2006; Razavi and Hosseinzadeh 2019; Zygmunt et al. 2010) and the polymethoxyflavones (Gao et al. 2018; R. W. Li et al. 2006; Onda et al. 2013) are the most studied compounds. According to in vitro and animal studies, citrus flavanones exert their anti-diabetic activity via several mechanisms; inhibition of gut glucose transporters, increase in insulin secretion, increased insulin sensitivity in target tissues and reduced hepatic glucose production and output (Ahmed et al. 2017; Erlund et al. 2001; Jung et al. 2004, 2006; Kerimi, Gauer, et al. 2019).

(Poly)phenols can act either within the gut and/or after absorption on different tissues, including the pancreas, liver, muscle, and adipose tissues, and these two processes are mostly independent. In the gut, modulation of glycemic response is possible mainly by affecting starch digestion and absorption (Sun and Miao 2020; Williamson 2013). After absorption, mechanisms include modulation of hepatic glucose metabolism and increase insulin secretion and sensitivity. Understanding the mechanism of action of (poly)phenols is crucial for the correct and appropriate design of human studies. For example, it is not worthwhile to perform a short study (≤2 hr) for a compound that takes longer than 2 hr to reach the circulation and acts mainly via post-absorptive mechanisms. The majority of the data on the mechanism of action of (poly)phenols originate from in vitro and animal studies. Hence, the following review paper mainly focuses on unraveling the anti-diabetic mechanism of citrus (poly)phenols by critically analyzing the experimental data, and also highlights approaches for future research.

Citrus (poly)phenols

Citrus fruits are rich in flavonoids, especially which constitute 90% of the (poly)phenols in citrus fruits, mostly as glycosides of hesperetin 8 and naringenin 9 (De Ancos et al. 2017; Ballistreri et al. 2019; Khan, Zill-E-Huma, and Dangles 2014). Citrus fruits are the main food source of hesperidin 1 and naringin 6 (Rampersaud and Valim 2017) and these compounds have not been detected in apples, berries (strawberry, bilberry, and cranberry), black grape nor sour cherries (Díaz-García et al. 2013). The (poly)phenol content and profile of citrus fruits depend mainly on species and variety. For example, hesperidin 1 is the main flavanone found in oranges, mandarins, lemons, and lime, while grapefruits contain naringin 6 in high concentrations (Table 1), contributing to the bitter taste (Díaz-García et al. 2013; Gattuso et al. 2007; Rampersaud and Valim 2017; Turner and Burri 2013). Apart from the flavanones, another group of compounds found almost exclusively in citrus fruits is the polymethoxyflavones (PMFs), particularly concentrated in the peel and flavedo of the fruit (Gao et al. 2018). Tangeretin 12 and nobiletin 13, the two most abundant PMFs found widely in citrus fruits, have been extensively studied owing to their distinct pharmacological properties (Gao et al. 2018; R. W. Li et al. 2006). Table 1 lists the (poly)phenol composition of orange, grapefruit, and lime juices, and the chemical structures of some of the predominant flavanones and polymethoxyflavones (PMF) in citrus fruits are illustrated in Figure 1.

Figure 1. Chemical structures of commonly found (poly)phenols in citrus fruits.

Table 1. (Poly)phenol composition of citrus juices.

	Citrus sinensis	Citrus paradisi	Citrus limon
Flavanones
Hesperidin 1	3.51–55.2	0.25–1.79	3.84–41
Narirutin 5	0.55–14.2	2.5–17.0	–
Naringin 6	–	4.5–60.2	–
Neohesperidin 4	–	0.67–1.58	–
Didymin 7	0.8–3.10	0.27–0.33	–
Eriocitrin 10	0.11–0.67	0.27–0.54	1.67–39.1
Neoeriocitrin	0.59	0.30–0.33	–
Poncirin	0.49–1.59	0.85–1.58	–
Flavones and flavonols
6,8-di-C-Glu-apigenin 11	4.15–8	–	1.0–1.45
6,8-di-C-Glu-diosmetin	0.25–0.45	–	4.05–5.8
Rhoifolin	0.05	0.28	–
Isorhoifolin	0.07	–	–
Diosmin	0.09	–	0.51–5.1
Neodiosmin	0.08	–	–
Rutin	–	3.26	–
Luteolin-7-O-rutinoside	–	–	1.5–6.5
Polymethoxyflavones
Heptamethoxyflavone	0.04–0.12	0.01–0.11	–
Nobiletin 13	0.19–0.46	0.12–0.17	–
Sinensetin	0.37	–	–
Tangeretin 12	0.01–0.07	0.12	–
Aglycones
Taxifolin	0.03	0.16	–
Acacetin	0.03	–	–
Hesperetin 8	–	0.74	–
Naringenin 9	–	0.98–8.0	–
Quercetin	–	0.17–0.21	–
Luteolin	–	–	0.08

Data represent values extracted from different studies for the juices (Gattuso et al. 2007). All values are expressed as mg (poly)phenol/100 mL juice.

Citrus fruit juices also contain flavone and flavonol glycosides of apigenin, luteolin, tamarixetin, isorhamnetin, kaempferol, and quercetin (Abad-García et al. 2012; Díaz-García et al. 2013; Gattuso et al. 2007; Sun et al. 2019; Tomás-Navarro, Vallejo, and Tomás-Barberán 2014). Considerable amounts of some flavones have been reported in C. limon (diosmetin 6,8-di-C-glucoside − 4.95 mg/100 mL, apigenin 6,8-di-C-glucoside (vicenin-2) − 1.17 mg/100 mL) and in C. sinensis (vicenin-2 − 5.72 mg/100 mL) juices (Gattuso et al. 2007; Tomás-Navarro, Vallejo, and Tomás-Barberán 2014). Phenolic acids such as 3′,4′-dihydroxycinnamic acid (caffeic acid), 4-hydroxy-3-methoxybenzoic acid (vanillic acid), 4′-hydroxycinnamic acid (p-coumaric acid), 3,4-dihydroxybenzoic acid (protocatechuic acid), 4′-hydroxy-3′-methoxycinnamic acid (ferulic acid), 4-hydroxy-3,5-dimethoxy-cinnamic acid (sinapic acid), and p-hydroxybenzoic acids were also detected in seven citrus varieties (Sun et al. 2019). The presence of chlorogenic and gallic (3,4,5-trihydroxybenzoic) acids have also been reported (Kelebek et al. 2009; Sdiri et al. 2012), but anthocyanins are rare and limited to some specific pigmented varieties of sweet and blood oranges and lemons (Ballistreri et al. 2019; Fallico et al. 2017).

Data from in vitro and animal studies

Possible antidiabetic activity of citrus (poly)phenols before absorption (in the gut)

In general, within the gut, the (poly)phenols mainly exert any hypoglycemic effects by attenuating carbohydrate digestion and absorption. They reduce starch digestion by inhibiting the starch digesting enzymes and blunt intestinal glucose absorption by either inhibiting the glucose transporters or by affecting the gene expression of these transporters (Kerimi, Nyambe-Silavwe, et al. 2019; Manzano and Williamson 2010; Nyambe-Silavwe et al. 2015; Lo Piparo et al. 2008; Rasouli et al. 2017; Schulze et al. 2014; Tadera et al. 2006; Villa-Rodriguez et al. 2017; Wu and Tian 2017). Inhibition of both the starch hydrolyzing enzymes and transporters in the gut slows down the release of glucose into the circulation thereby preventing hyperglycemic episodes, a risk factor for the development of type 2 diabetes (Bhupathiraju et al. 2014; Greenwood et al. 2013; Livesey et al. 2008, 2019). Figure 2 illustrates the overall summary of results extracted from studies published so far on the effect of citrus (poly)phenols on starch digesting enzymes and sugar transporters using both in vitro and animal studies.

Figure 2. Potential inhibitory mechanisms of citrus (poly)phenols on starch digestion and absorption within the gut. SGLT1 (Sodium-dependent glucose cotransporter 1), GLUT2 (Glucose transporter 1), and GLUT5 (Glucose transporter 5) are sugar transporters located in the brush border membrane.

Effect on starch hydrolyzing enzymes

Salivary and pancreatic α-amylases, and the intestinal brush border α-glucosidases (maltase-glucoamylase and sucrase-isomaltase), are the key enzymes involved in digesting dietary starch in humans (Butterworth, Warren, and Ellis 2011; Goodman 2010; Lin, Lee, and Chang 2016; Pyner, Nyambe-Silavwe, and Williamson 2017; Visvanathan et al. 2019). α-Amylase hydrolyzes starch to maltooligosaccharides by acting upon the internal α-1,4-glycosidic linkages. Since α-amylase is an endo-acting enzyme, the products are maltodextrins with no free glucose (Butterworth, Warren, and Ellis 2011; Visvanathan et al. 2019, Visvanathan, Houghton, and Williamson 2021). The digested maltodextrins are then concomitantly transformed into glucose by the action of intestinal brush border α-glucosidases (Kerimi, Nyambe-Silavwe, et al. 2019; Lin, Lee, and Chang 2016; Pyner, Nyambe-Silavwe, and Williamson 2017). Thus, inhibition of these two enzymes aid in attenuating postprandial glucose peaks after a meal. The inhibitory activity of dietary (poly)phenols on starch digesting enzymes is well documented (Fei et al. 2014; Nyambe-Silavwe et al. 2015; Pyner, Nyambe-Silavwe, and Williamson 2017; Tadera et al. 2006; Visvanathan, Houghton, and Williamson 2021). However, the inhibitory activity of citrus (poly)phenols on these enzymes is not yet clearly defined and studies done using enzymes from human sources are rare (Table 2). This makes interpretation difficult, as enzymes from other sources such as microbial enzymes have very little sequence homology or similarity in inhibitory patterns to human enzymes, and mammalian enzymes show some similarities but also some important differences. For example, Priscilla et al. (2014) clearly show the influence of enzyme sources on the inhibitor by (poly)phenols. In that work, the inhibitory potential of naringenin 9 toward α-glucosidase from yeast and rat intestine was studied. Naringenin 9 showed dramatically better inhibitory potential against yeast α-glucosidase (IC₅₀ = 6.51 µM) compared to rat intestinal α-glucosidase (IC₅₀ = 384 µM). The opposite was observed with acarbose (IC₅₀ = 49.65 and 2.45 µM for the yeast and mammalian α-glucosidases, respectively). Similarly, a comparison between human and rat intestinal α-glucosidases revealed EGCG to show differential inhibition of enzymes from different sources (Pyner, Nyambe-Silavwe, and Williamson 2017).

Table 2. In vitro studies on citrus compounds and the markers of type 2 diabetes.

Mechanism	Reference	Test model/ Source	Test sample	Test concen-tration	Positive control	Outcome	Physiological relevance		Comments
							Test sample	Test/ active dose
Inhibition of starch digesting enzymes	(Casacchia et al. 2019)	Human pancreatic α-amylase	Clementine; Tarocco tetraploids; cross-bred hybrid Tacle	0.05–400 μg/mL	Acarbose	IC50 ranged from 44.3 ± 0.7 to 99.66 ± 0.36 µg/mL. Acarbose IC50 = 36.5 ± 0.3 μg/mL.	√	√	–
	(Uddin et al. 2014)	α-amylase (unknown source)	Citrus macroptera Montr. fruit extract	500–2000 μg/mL	Acarbose	Citrus extract inhibited α-amylase (IC50 = 3.64 mg/mL); acarbose (IC50 = 0.91 mg/mL).	√	√	Enzyme source not known.
	(Oboh and Ademosun 2011)	Porcine pancreatic α-amylase and yeast α-glucosidase	Shaddock (Citrus maxima) peel extract	80–320 µg/mL	–	Moderate α-amylase and strong α-glucosidase inhibition.	√	√	α-glucosidase from yeast not relevant to human enzyme; porcine α-amylase somewhat relevant.
	(Alu’datt et al. 2017)	α-amylase (unknown source) and yeast α-glucosidase	Shamouti and red blood orange; white grapefruit; lemon; white pummelo; clementine	–	–	Exhibited poor/no α-amylase and strong α-glucosidase inhibitory activity.	√	–	α-glucosidase from yeast not relevant to human enzyme.
	(Kim, Shin, and Jang 2009)	Rat intestinal α-glucosidase and porcine pancreatic α-amylase	Citrus grandis Osbeck (CGO) extract; CGO aglycones; Aspergillus saitoi-treated CGO extract	–	–	CGO extract showed poor α-glucosidase inhibition and no α-amylase inhibition. A. saitoi-treated CGO extract inhibited the enzymes >50%; aglycones showed moderate inhibition.	√	–	Enzymes from non-human sources have only partial relevance to human enzymes.
	(Menichini et al. 2011)	Porcine pancreatic α-amylase and yeast α-glucosidase	Citrus medica L. extract	5–1000 μg/mL	Acarbose	Weak inhibition: α-amylase and α-glucosidase (∼8.5- and ∼16-fold lower than acarbose, respectively).	√	√	α-glucosidase from yeast not relevant to human enzyme; porcine α-amylase somewhat relevant.
	(de la Garza et al. 2013)	Porcine pancreatic α-amylase and yeast α-glucosidase	Grapefruit extract	20–1500 μg/mL	Acarbose	Poor inhibition of α-amylase and α-glucosidase (IC50 = 420 and 410 µg/mL, respectively). Acarbose (IC50 = 4.1 and 6.9 µg/mL)	√	√	α-glucosidase from yeast not relevant to human enzyme; porcine α-amylase somewhat relevant.
	(Gong et al. 2017)	Yeast α-glucosidase	Hesperetin 8	0–2500 μM	–	Inhibited yeast α-glucosidase with IC50 value = 380 µM.	√	X	α-glucosidase from yeast not relevant to human enzyme. Hesperetin 8 used at much higher concentration than physiological.
	(Kerimi, Gauer, et al. 2019)	Rat intestinal sucrase	Hesperidin 1	0–1000 μM	Acarbose	No inhibition.	√	√	Rat enzyme only somewhat relevant to human.
	(Zheng, Lu, and Xing 2021)	Yeast α-glucosidase	Hesperidin 1; hesperidin hydrolysates (acid hydrolysis); hesperetin 8	60–10000 μg/mL	–	Hesperidin hydrolysates exhibited strong α-glucosidase inhibition (IC50 = 83 μg/mL); hesperetin 8 moderate effect. Hesperidin 1 very weak (<15% at 10 mg/mL).	√	√	α-glucosidase from yeast not relevant to human enzyme. Only trace levels of the aglycones are present naturally in the fruit; only a very small amount is produced after deglycosylation in the small intestine.
	(Shen, Xu, and Lu 2012)	Porcine pancreatic α-amylase and yeast α-glucosidase	Hesperidin 1; naringin 6; neo-hesperidin 4; nobiletin 13	12.5–160 μM	Acarbose	Weak inhibition	√	√	α-glucosidase from yeast not relevant to human enzyme; porcine α-amylase somewhat relevant.
	(Priscilla et al. 2014)	Yeast and rat intestinal α-glucosidase	Naringenin 9	3.15–794 μM	Acarbose	Naringenin 9 inhibited yeast and rat intestinal α-glucosidase with an IC50 value of 6.51 and 384 µM, respectively (Acarbose 49.7 and 2.45 µM, respectively).	√	√	α-glucosidase from yeast not relevant to human enzyme; rat α-glucosidase somewhat relevant. Only trace levels of the aglycones are present naturally in the fruit; only a very small amount is produced after deglycosylation in the small intestine.
	(Ali et al. 2019)	Yeast α-glucosidase	Didymin 7	0–250 µM	Acarbose	Didymin 7 (IC50 = 48.8 ± 1.0 μM) was 2.6 times higher than acarbose (IC50 = 129.9 ± 0.8 μM).	√	√	α-glucosidase from yeast not relevant to human enzyme.
Glucose transporters	(de la Garza et al. 2013)	Caco-2 cells	Grapefruit extract	480 µg/mL	Phloridzin (500 µM)	Grapefruit extract inhibited SGLT1-mediated methyl glucoside uptake by 29% (Phloridzin − 56%).	√	√	–
	(Kerimi, Gauer, et al. 2019)	Caco-2/TC7 cell and Xenopus laevis oocytes expressing human GLUT2 or GLUT5	Hesperidin 1; hesperetin 8; orange juice	80–1600 μM	–	Hesperidin 1 and orange juice moderately inhibited the glucose transporters in Caco-2/TC7 cells.	√	√	–
	(J. M. Li et al. 2006)	Rat everted intestinal sleeves	Naringenin 9	10–500 µM	Phloridzin	Naringenin 9 strongly inhibited glucose uptake (IC50 = 2.4 µM); phloridzin (IC50 = 1.9 µM).	√	√	–
	(Johnston et al. 2005)	Caco-2/TC7 cells	Naringin 6; neohesperidin dihydrochalcone	100 μM	–	Neohesperidin dihydrochalcone inhibited glucose absorption (∼50%) by inhibiting SGLT1. Naringin was weaker (<20%).	√	x (neo-hesper-idin)	The active neohesperidin 4 concentration is far higher than found by ingesting citrus fruits (<26 μM).
	(Kwon et al. 2007)	Xenopus laevis oocytes expressing GLUT2 or GLUT5	Hesperetin 8; naringenin 9; naringin 6	0–300 μM	–	Hesperetin 8 inhibited deoxy-D-glucose and fructose absorption with IC50 = 48.6 and 66.5 μM, respectively. Naringenin 9 poor inhibition (IC50 = >300 μM); naringin 6, no inhibition	√	√	Only trace levels of the aglycones are present naturally in the fruit; only a very small amount is produced after deglycosylation in the small intestine.
Insulin sensitivity	(Horiba et al. 2010)	3T3-L1 adipocytes	Chronic naringenin chalcone treatment (8 days)	25–100 μM	–	Naringenin chalcone (≥25 μM) ↑ adiponectin mRNA and protein secretion via PPAR-γ.	√	x	Active dose is far above the level of naringenin metabolites (21–26) in plasma after consuming citrus fruits (<6 μM).
	(Rebello et al. 2019)	Human white adipocyte (hADSC) and abdominal subcutaneous adipose tissue (pWAT)	Chronic naringenin 9 treatment hADSC − 3 or 7 days pWAT − 7 or 14 days	8 µM	–	In both cell lines, ↑ mRNA expression of GLUT4, adiponectin, and ChREBP.	√	√	–
	(Murugesan et al. 2020)	Human subcutaneous adipocyte	Chronic naringenin 9 treatment for 2 days	8 µM	–	No effect on GLUT4 mRNA expression.	√	√	–
	(Zhao et al., 2020)	HTR-8/SVneo and HUVEC cells	Chronic naringenin 9 treatment for 2 days	50 µM	–	Naringenin 9 ↑ insulin-stimulated glucose uptake by reducing miR-140 expression and ↑ insulin signaling pathway.	√	x	Active dose is far above the level of naringenin metabolites (21–26) found in plasma after consuming citrus fruits (<6 μM).
	(Richard et al. 2013)	3T3-L1 adipocytes and sub-cutaneous human adipocytes	Naringenin 9	44–184 µM	–	↓ differentiation of preadipocytes and ↑ insulin resistance by ↓ IRS-1 tyrosine phosphorylation in 3T3-L1 adipocytes. ↓ adiponectin expression in both human (92 µM) and murine mature adipocytes (184 µM).	√	x	Active doses are far above the level of naringenin metabolites (21–26) found in plasma after consuming citrus fruits (<6 μM).
	(Harmon and Patel 2003)	3T3-L1 adipocytes	Naringenin 9	25–100 μM	–	Naringenin 9 (≥25 μM) inhibited insulin-stimulated glucose uptake by inhibiting PI3K.	√	x	Active doses are far above the physiological plasma concentration of naringenin metabolites (21–26) (<6 μM).
	(Yoshida et al. 2013)	3T3-L1 adipocytes	Naringenin 9	1–100 µM		Naringenin 9 > 10 µM ↓ TLR2 mRNA and protein expression during adipocyte differentiation, with no effect in pre- and matured adipocytes.	√	√	–
	(Yang et al. 2013)	MDA-MB-231 breast cancer cells	Hesperetin 8	5–100 µM	–	Hesperetin 8 ↓ basal (≥50 µM) and insulin-stimulated glucose uptake (≥100 µM), down-regulated GLUT1, ↓ insulin-induced translocation of GLUT4, suppressed phosphorylation of insulin receptor-β subunit and Akt.	√	x	Active doses are far above the reported physiological plasma concentrations of hesperetin metabolites (14–20) (<2.2 μM).
	(Claussnitzer et al. 2011)	3T3-L1 adipocytes and mature human adipocytes	Hesperidin 1; naringin 6; naringenin 9	20 µM	–	Naringenin 9 ↓ basal (no insulin) and insulin-induced glucose uptake in both cells. Hesperidin 1 ↓ insulin-induced glucose absorption in 3T3-L1 adipocytes; naringin 6, no effect.	x (hesperidin 1, naringin 6)	x	Hesperidin 1 and naringin 6 do not reach circulation in their original forms so physiologically not relevant. Active dose of naringenin 9 is far above plasma concentrations of naringenin metabolites (21–26) (<6 μM).
Mechanism	Reference	Test model/ Source	Test sample	Test concen-tration	Positive control	Outcome	Physiological relevance		Comments
							Test sample	Test/ active dose
	(Liu et al. 2008)	3T3-L1 adipocytes	Hesperetin 8; naringenin 9	20 − 160 µM	Rosiglitazone (10 μM)	↑ adiponectin transcription (80 µM) and PPAR-γ controlled luciferase expression (≥20 µM) in a dose-dependent manner.	√	x	Active doses are far above the reported plasma concentrations of flavanone metabolites (14–26) (<2.2 and 6 μM for hesperetin 8 and naringenin 9, respectively).
	(Park 1999)	Myelocytic U937 cells	Hesperetin 8; naringenin 9; naringin 6	0–100 μM	–	Hesperetin 8 and naringenin 9 significantly inhibited glucose uptake (IC50 = ∼50 µM) in U937 cells in a dose-dependent manner, while naringin 6 did not affect uptake.	x (naringin)	x	Naringin 6 does not reach circulation in its original forms so physiologically not relevant. The active concentrations of the aglycones are higher than the reported physiological plasma concentration of the respective metabolites (14–26) (<2.2 and 6 μM for hesperetin 8 and naringenin 9, respectively).
	(Nomura et al. 2008)	MC3T3-G2/PA6 adipocytes	Hesperetin 8; naringenin 9; hesperidin 1	10 μM	–	No effect on deoxy-D-glucose uptake	x (hesperidin 1)	√	Hesperidin 1 is not absorbed in its original form.
	(Onda et al. 2013)	3T3-F442A adipocytes	Tangeretin 12; nobiletin 13	5–50 µM	–	Tangeretin 12 and nobiletin 13 (≥5 µM) ↑ the uptake of deoxy-D-[1-^3H]- glucose dose-dependently by modulating the PI3K/protein kinase A pathways.	√	√	–
	(Kim et al. 2012)	C2C12 myotubes	Tangeretin 12	0–200 μM	–	Tangeretin 12 (>100 μM) significantly ↑ glucose uptake and phosphorylation of AMPK.	√	x	The active concentration is not achieved under normal physiological conditions (tangeretin 12 content in the edible portion of citrus fruit <5 μM).
	(Miyata et al. 2011)	3T3-L1 adipocytes	Tangeretin 12; nobiletin 13	16–128 µM	Pioglitazone (for adiponectin test)	Both flavonoids ↑ secretion of adiponectin (16–64 µM), and ↓ secretion of MCP-1 (128 µM). Nobiletin 13 at 128 µM ↓ secretion of resistin.	√	x	The active concentrations are not achieved under normal physiological conditions (tangeretin 12 and nobiletin 13 content in edible portion of citrus fruit <11 μM).
	(Saito, Abe, and Sekiya 2007)	3T3-L1 adipocytes	Nobiletin 13	60 μM	–	Nobiletin 13 enhanced differentiation, upregulated adiponectin mRNA levels, and ↑ lipolysis.	√	x	The active concentration is not achievable under normal physiological conditions (nobiletin 13 content in the edible portion of citrus fruit <11 μM).
	(Kanda et al. 2012)	3T3-L1 adipocytes and mouse embryonic fibroblasts	Nobiletin 13	1–100 µM	–	Suppressed adipocyte differentiation (≥1 µM) in 3T3-L1 preadipocytes and mouse embryonic fibroblasts.	√	√	–
	(Kunimasa et al. 2009)	ST-13 adipocytes	Nobiletin 13	10 μM	Troglitazone	Nobiletin 13 induced differentiation of preadipocytes into mature adipocytes; ↑ adiponectin (3.4-fold).	√	√	–
	(Ali et al. 2019)	Insulin-resistant HepG2 cells	Didymin 7	10–30 µM	Metformin	Didymin 7 (≥10 µM) improved insulin-dependent glucose uptake by down- regulating PTP1B, leading to ↑ phosphorylation of IRS-1, PI3K, Akt, and GSK-3. ↓ PEPCK and G6Pase expression.	x	√	Didymin 7 is not absorbed in its original form and so will not reach the liver in this form.
	(Gamo et al. 2014)	3T3-L1 adipocytes	Hesperetin-7- glucuronide 14; hesperetin-3′- glucuronide 16	250 µM	Troglitazone (10 µM)	Enhanced adipocyte differentiation by activating PPAR-γ; comparable to troglitazone.	√	x	The active concentration is not achieved under normal physiological conditions (maximum reported 2.2 μM).
	(Nguyen‐Ngo, Willcox, and Lappas 2019)	TNF-α-stimulated human skeletal muscle tissue explant	Naringenin 9	400 μM	–	Naringenin 9 ↑ insulin-stimulated glucose uptake in TNF-α stimulated gestational diabetes model skeletal muscle cells.	√	x	The active concentration is not achieved under normal physiological conditions (maximum reported 6 μM).
	(Li et al. 2019)	TNF-α-stimulated C2Cl2 myoblasts	Naringenin 9	184 µM	–	Naringenin 9 significantly ↑ GLUT4 translocation and glucose uptake in TNF-stimulated insulin-resistant muscle cells.	√	x	The active concentration is not achieved under normal physiological conditions (maximum reported 6 μM).
Insulin secretion	(Gupta et al. 2018)	Human recombinant DPP-4	Rutin; eriocitrin 10; eriodictyol; naringenin 9; naringin 6; hesperidin 1; hesperetin 8	–	Sitagliptin; saxagliptin; vildagliptin	Poor DPP-4 inhibition (IC50 = 485 to 5700 µM); commercial drugs (IC50 values sitagliptin − 0.684 µM, saxagliptin − 0.707 µM, and vildagliptin − 2.286 µM).	x (naringin 6, hesperidin 1, rutin, eriocitrin 10)	x	Naringin 6, hesperidin 1, eriocitrin 10 and rutin are not absorbed in their original form. The active concentrations are not achieved under normal physiological conditions.
	(Fan et al. 2013)	Porcine kidney DPP-4	Hesperetin 8; naringenin 9; eriocitrin 10	–	Diprotin A	Hesperetin 8 and naringenin 9 showed potential DPP-4 inhibition (IC50 = 0.28 and 0.24 µM, respectively) compared to positive control (IC50 = 4.21 µM).	x (eriocitrin 10)	√	Study used enzyme from non-human sources. Eriocitrin 10 is not absorbed in its original form.
	(Parmar et al. 2012)	DPP-4 (unknown source)	Naringin 6	9 µM	Sitagliptin (9 µM)	Naringin exhibited potent DPP-4 inhibitory activity (∼69%), comparable to sitagliptin (∼60%).	x	x	Naringin is not absorbed in its original form.
	(Xing, Yang, and Wu 2016)	Human embryonic stem cells-derived pancreatic endoderm	Naringenin 9	367 µM	–	Naringenin 9 ↑ glucose-stimulated insulin secretion.	√	x	The active concentration is not achieved under normal physiological conditions (maximum reported 6 μM).
	(Bhattacharya et al. 2014)	INS-1E cells	Naringenin 9	10^−10, 10^−8, 10^−6 M	–	Chronic and acute naringenin 9 exposure ↑ glucose-stimulated insulin secretion in a dose-dependent manner. Naringenin 9 ↑ GLUT2, GK, Ins1, Ins2, Akt1, and Akt2 mRNA expression.	√	√	–
Insulin mimetic activity	(Borradaile, de Dreu, and Huff 2003)	HepG2 cells	Naringenin 9	200 μM	Insulin (100 nM)	Naringenin 9 ↑ LDL receptor and SREBP-1 expression and ↓ apoB secretion via activating PI3K independent of insulin.	√	x	The active concentration is not achieved under normal physiological conditions (maximum reported 6 μM).
	(Allister et al. 2005)	HepG2 cells	Naringenin 9	100 μM	Insulin (100 nM)	Naringenin 9 ↓ MTP expression, resulting in inhibition of apoB100 secretion.	√	x	The active concentration is not achieved under normal physiological conditions (maximum reported 6 μM).
	(Lim, Soh, and Kuppusamy 2008)	Rat adipocyte primary cells	Naringenin 9	0.01–100 nM	Insulin (1 and 100 µM)	Naringenin 9 (≥0.01 µM) inhibited epinephrine-induced lipolysis by ∼50%, boosted (1 µM) antilipolytic effect of insulin, ↑ glucose uptake in a dose-dependent manner.	√	√	–
	(Zygmunt et al. 2010)	L6 myotubes	Naringenin 9	10–150 μM	Insulin (100 nM)	Naringenin 9 at >50 μM ↑ glucose uptake, independent of insulin via AMPK pathway.	√	x	The active concentration is not achieved under normal physiological conditions (maximum reported 6 μM).
Hepatic glucose homeostasis	(do Nascimento et al. 2018)	Isolated perfused rat liver	Naringenin 9; hesperetin 8; hesperidin 1	100–300 μM	–	All (300 μM) stimulated glycogenolysis and glycolysis. Naringenin 9 and hesperetin 8 inhibited G6P (≥100 µM) and GK (≥50 µM), and mitochondrial respiration linked to ADP phosphorylation. Hesperetin 8 and naringenin 9 (300 μM) inhibited glucose uptake.	x (hesperidin)	x	Hesperidin is not absorbed in its original form. The active concentrations are not achieved under normal physiological conditions (<2.2 and 6 μM for hesperetin 8 and naringenin 9, respectively).
	(Shen, Xu, and Lu 2012)	HepG2 cells	Hesperidin 1; neohesperidin 4; naringin 6; nobiletin 13	10 and 50 μM	Metformin	The flavonoids (≥10 µM) ↑ glucose consumption, glycogen level, hepatic GK activity, and ↓ G6P activity.	x (hesperidin 1, neo-hesperidin 4, naringin 6)	√	Hesperidin 1, neohesperidin 4 and naringin 6 are not absorbed in their original form.
	(Constantin et al. 2014)	Isolated perfused rat liver	Hesperidin 1; hesperetin 8; naringenin 9	25–300 μM	–	Hesperetin 8 and naringenin 9 inhibited hepatic gluconeogenesis from lactate, pyruvate, and alanine, mainly by inhibiting pyruvate carboxylase (IC50 = 134.2 and 143.5 μM, respectively).	x (hesperidin 1)	x	Hesperidin 1 is not absorbed in its original form. The active concentrations are not achieved under normal physiological conditions (< 2.2 and 6 μM for hesperetin 8 and naringenin 9, respectively).
	(Purushotham, Tian, and Belury 2009)	Fao hepatoma cells	Naringenin 9; naringin 6	6–100 µM	Metformin (1 mM)	Naringenin 9 (≥25 μM) significantly ↓ hepatic glucose production in a dose-dependent manner, but not naringin 6.	x (naringin 6)	x	Naringin 6 is not absorbed in its original form. The active concentrations not achieved under normal physiological conditions (maximum for naringenin 9 is 6 μM).

Physiologically relevant test compound and dose (√), and physiologically irrelevant test compound and dose (x). Compounds shown within parenthesis in the column with the (x) sign indicate the corresponding compound.

Abbreviations: ↓ – Decreased; ↑ – Increased; ADP – Adenosine diphosphate; Akt – Protein kinase B; AMPK – 5’ adenosine monophosphate-activated protein kinase; apoB – Apolipoprotein B; ChREBP – Carbohydrate response element-binding protein; DMSO – Dimethyl sulfoxide; DPP-4 – Dipeptidyl-peptidase 4; G6P – Glucose 6-phosphatase; GK – Glucokinase; GLUT1 – Glucose transporter 1; GLUT4 – Glucose transporter 4; GSK-3 – Glycogen synthase kinase-3; Ins1 – Insulin 1;Ins2 – Insulin 2; IRS-1 – Insulin receptor substrate; LDL – Low-density lipoprotein; MCP-1 – Monocyte chemoattractant protein; miR-140 – MicroRNA-140; mRNA – Messenger ribonucleic acid; MTP – Microsomal triglyceride transfer protein; PI3-K – Phosphatidylinositol 3 kinase; PPAR-γ – Peroxisome proliferator-activated receptor gamma; PTP1B – protein tyrosine phosphatase 1B; SREBP-1 – Sterol regulatory element-binding protein; TLR-2 – Toll-like receptor 2; TNF-α – Tumor necrosis factor-α.

Few studies have reported the effect of pure citrus extracts or the constituent (poly)phenols on these enzymes (Table 2). According to Kerimi, Gauer, et al. (2019), hesperidin 1 did not exhibit any potential inhibition toward rat intestinal sucrase and human α-amylase. Similarly, hesperidin 1, naringin 6, neohesperidin 4, and nobiletin 13 showed weak inhibition against the digestive enzymes but did show good starch binding ability, which significantly delayed starch digestion (Shen, Xu, and Lu 2012). All four compounds exhibited poor α-glucosidase (yeast) inhibitory activity, and for α-amylase (porcine), nobiletin 13 showed stronger inhibition compared to the other three citrus flavonoids. However, the inhibitory potential was below 30% even at the highest concentration tested (160 µM). Dangyuja (Citrus grandis Osbeck) water extract, containing mainly neohesperidin 4 (160 mg/g extract) and naringin 6 (110 mg/g extract), showed weak inhibition toward rat intestinal α-glucosidase and no inhibitory activity on porcine pancreatic α-amylase (Kim, Shin, and Jang 2009). The aglycones hesperetin 8 and naringenin 9, obtained from hesperidinase- and naringinase-treated extract, improved the inhibitory potential moderately (Kim, Shin, and Jang 2009). In a study done using human pancreatic α-amylase, an extract from a cross-bred citrus variety (Clementine and Tarocco tetraploids) strongly inhibited α-amylase, close to the activity of the commercially used anti-diabetic drug acarbose (Table 2) (Casacchia et al. 2019). However, more studies using human enzymes are required but, so far, published studies do not indicate promising starch digesting enzyme inhibitory effect from citrus (poly)phenols at physiologically relevant concentrations.

Effect on the glucose transporters

Few studies have reported the possible inhibitory effect of citrus components on glucose absorption in the gut (Tables 2 and 3). Hesperidin 1, hesperetin 8, and whole orange juice inhibited the glucose transporters, GLUT2, SGLT1, and GLUT5 in Caco-2/TC7 cell monolayers (Kerimi, Gauer, et al. 2019). Glucose is primarily absorbed by SGLT1 and GLUT2 transporters in the small intestine, whereas GLUT5 is mainly involved in fructose absorption. Orange juice reduced glucose transport by 43% in differentiated Caco2/TC7 cells, compared to a sugar-matched control (Kerimi, Gauer, et al. 2019). Hesperidin 1 concentrations of 80 and 800 µM significantly inhibited glucose transport in Caco-2/TC7 cells but at very high concentrations well above that observed in orange juice (1600 µM), hesperidin 1 appeared to enhance glucose transport (Kerimi, Gauer, et al. 2019). The solubility of hesperidin 1 in aqueous systems is very low with maximum solubility in pure water not exceeding 40 µM (Tommasini et al. 2005). The authors state (Kerimi, Gauer, et al. 2019) that hesperidin 1 is soluble up to a concentration of 800 µM when first dissolved in DMSO and then introduced into orange juice, so presumably some factor keeps hesperidin in an apparent solution-like state. Solubility of hesperidin 1 at higher concentrations (>800 µM) is poor and so some effects could be due to a suspension of hesperidin at such high concentration. This could explain why the 1600 µM is less active than the 800 µM in some systems, for example. Solubility will certainly affect activity given that (poly)phenol solubility is a major factor determining bioavailability and bioactivity (Cao et al. 2018; Nguyen et al. 2015; Vallejo et al. 2010).

Table 3. In vivo studies using animal models on citrus compounds and type 2 diabetes.

Model	Treatment	Challenge	Test duration and route of administration	Test sample and dose	Outcome	References
C57BL/KsJ-db/db type 2 diabetic mice	Control Vs. hesperidin 1 Vs. naringin 6	–	5 weeks, oral	200 mg hesperidin or naringin/kg diet (0.02%)	Hesperidin and naringin ↓ blood glucose, ↑ plasma insulin, C-peptide, leptin, and hepatic glycogen levels, and hepatic GK activity. Naringin also ↓ hepatic G6P and PEPCK activity.	(Jung et al. 2004)
C57BL/KsJ-db/db mice with leptin receptor mutation	Control Vs. hesperidin 1 Vs. naringin 6	–	5 weeks, oral	200 mg hesperidin or naringin/kg diet (0.02%)	Naringin and hesperidin ↑ hepatic GK mRNA, adipocyte GLUT4, hepatic and adipocyte PPARγ expression, and ↓ hepatic GLUT2 protein. Naringin also ↓ mRNA expression of PEPCK and G6P.	(Jung et al. 2006)
Goto-Kakizaki rats with type 2 diabetes	Control Vs. hesperidin 1 Vs. cyclodextrin (CD)-clathrated hesperetin	–	4 weeks, oral	10,000 mg hesperidin 1/kg diet (1%) and 46 g CD-hesperetin /kg diet (4.6%)	Hesperidin 1 and CD-hesperetin normalized blood glucose levels by modulating activities of GK (↑) and G6P (↓); changed PPARs, HMG-CoA reductase, LDL receptor mRNA.	(Akiyama et al. 2009)
STZ-induced marginal type 1 diabetic rats	Control Vs. hesperidin 1	–	4 weeks, oral	10,000 mg hesperidin/kg diet (1 %)	Hesperidin ↓ blood glucose and G6P activity, and ↑serum insulin, adiponectin, and GK activity.	(Akiyama et al. 2010)
STZ-induced diabetic rats	Grapefruit juice	Glucose (3 g/kg bw) intraperitoneal	60 days, oral	3 mL/kg bw grapefruit juice	Improved FBG and glucose intolerance. ↑ hepatic glycogen content and GK activity and ↓ G6P and PEPCK activities.	(Hayanga et al. 2016)
Sprague-Dawley female non-diabetic rats	Control (water) Vs. citrus extract (Citrus macroptera Montr) Vs. glibenclamide	Glucose (1 g/kg bw)	3 hr	500 and 1000 mg/kg bw citrus extract 5 mg /kg bw glibenclamide	Both doses significantly reduced blood glucose after 2 hr. No change was observed after 1 hr of glucose administration compared to the control.	(Uddin et al. 2014)
STZ/NA-induced type 2 diabetic rats	Control Vs. Navel orange peel hydroethanolic extract Vs. naringin 6 Vs. naringenin 9	–	4 weeks, oral	100 mg/kg bw	Improved impaired oral glucose tolerance, ↑ serum insulin and C-peptide levels, and ↓ hepatic glycogenolysis and glucose output. ↑ insulin receptor β-subunit, GLUT4, and adiponectin mRNA in adipose tissues.	(Ahmed et al. 2017)
Wistar albino male rats	Control Vs. naringin 6 Vs. sitagliptin	–	9 days, oral	40 mg/kg bw twice daily	Naringin ↓ serum DPP-4 activity (31%) and blood glucose (53%), and ↑ serum insulin (68%).	(Parmar et al. 2012)
C57BL/6J mice	Control Vs. clarified grapefruit juice Vs. naringin 6 Vs. metformin	High-fat diet	100 days, oral	Clarified grapefruit juice 25 and 50% per day mixed with drinking water Naringin (0.72 mg/day) Metformin (7.5 mg/day)	Grapefruit juice ↓ body weight, liver TG, FBG, and FBI. Naringin ↓ FBG and FBI, but not body weight.	(Chudnovskiy et al. 2014)
High-fat diet-induced obesity/diabetes mouse model	Control Vs. Citrus reticulata Blanco peel extract	High-fat diet	15 weeks, oral	2500 mg (0.25%) and 500 mg (0.5%) citrus extract/kg diet	Citrus extract improved glucose homeostasis, insulin sensitivity, and lipid metabolism.	(Guo et al. 2016)
Healthy Wistar rats	Control Vs. grapefruit extract	Starch or maltose or glucose (2 g/kg bw)	3 hr, oral gastric intubation	500 mg/kg bw grapefruit extract	Grapefruit extract did not significantly reduce postprandial hyperglycemia after maltose or glucose tolerance tests.	(de la Garza et al. 2013)
Insulin resistant Wistar rats	Control Vs. grapefruit extract	High-fat sucrose diet	5 weeks, oral	1000 mg/kg bw grapefruit extract	Grapefruit improved hyperinsulinemia (50%) and HOMA-IR (54%).	(de la Garza et al. 2013)
Fructose induced insulin-resistant rats	Control Vs. naringenin 9	High-fructose diet (60% fructose)	44 days, oral	50 mg naringenin 9/kg bw	Naringenin 9 restored fructose-induced insulin resistance. ↓ FBG, FBI, and HOMA-IR. Further ↑ hepatic and muscle glycogen level, hexokinase, pyruvate kinase activity and ↓ G6P, glycogen phosphorylase, and F16BP activity.	(Kannappan and Anuradha 2010)
C57BL/6J mice with metabolic syndrome	Control Vs. naringin 6	High-fat diet	20 weeks, oral	200 mg naringin/kg diet (0.02%)	Naringin ↓ FBG, FBI, HOMA-IR, serum leptin, TNF-α, hepatic G6P and PEPCK, and ↑ serum adiponectin.	(Pu et al. 2012)
Wistar rats	Control Vs. citrus hybrid Tacle^®	High-fat diet	12 weeks, oral	20 mg extract daily	Citrus intervention ↓ adiposity, LDL-C, TC, TG, and improved insulin sensitivity and glucose tolerance.	(Casacchia et al. 2019)
STZ-induced diabetic rats	Control Vs. tangeretin 12 Vs. glibenclamide	–	75 days, intragastric	25, 50, 100 mg tangeretin 12/kg bw 5 mg glibenclamide/kg bw	100 mg/kg bw tangeretin 12 treatment ↓ FBG, HbA1c and ↑ FBI, hepatic glycogen levels, and altered hepatic enzyme levels to normalcy, comparable to glibenclamide.	(Sundaram, Shanthi, and Sachdanandam 2014)
STZ-induced insulin-resistant diabetic rats	Control Vs. Citrus sinensis fruit peel extract Vs. metformin	High-fat diet	30 days, oral gavage	50 and 100 mg citrus extract/kg bw 100 mg metformin/kg bw	Citrus treatment (50 and 100 mg) ↓ FBG, FBI, and HOMA-IR levels and ↑ PPARγ, GLUT4, and insulin receptor mRNA and protein expression. Effects comparable to metformin.	(Sathiyabama et al. 2018)
C57BL/KsJ-db/db mice with type 2 diabetes	Control Vs. Citrus unshiu peel ethanol extract Vs. rosiglitazone	–	6 weeks, oral	20 mg citrus peel extract/g diet (2%) Rosiglitazone (0.001%)	Citrus extract ↓ weight gain, body fat mass, blood glucose level, hepatic PEPCK, MCP-1, TNF-α, and IL-6 level, and ↑ adiponectin and IL-10 levels.	(Park et al. 2013)
OFA rats with metabolic syndrome	Control Vs. hesperidin 1 and naringenin 9 supplemented biscuit	Obesogenic cafeteria diet	8 weeks, oral	40 mg hesperidin 1 and 20 mg naringenin 9/kg bw	The flavanone supplemented biscuit ↓ systolic blood pressure, serum TC and LDL-C, abdominal obesity, and improved insulin resistance.	(Mayneris-Perxachs et al. 2019)
Fructose-induced insulin resistant hamsters	Control Vs. polymethoxyflavone supplement	Fructose-rich diet (60% fructose)	4 weeks, oral	62.5 and 125 mg/kg bw polymethoxy flavone	PMF (125 mg) improved insulin resistance and glucose tolerance. ↓ expression of TNF-α, IFN-γ, IL-1β, and IL-6 and ↑ serum adiponectin, and hepatic PPARα and PPARγ protein expression.	(R. W. Li et al. 2006)
C57BL/6 J mice	Control Vs. tangeretin 12	High-fat diet	8 weeks, oral	200 mg tangerine/kg diet (0.02%)	Tangeretin 12 improved insulin resistance, ↓ leptin, resistin, adiponectin, MCP-1, and IL-6.	(Kim et al. 2012)
STZ-induced type 2 diabetic rats	Control Vs. naringenin 9 Vs. metformin	High-fat diet	6 weeks, intragastric	50 and 100 mg naringenin 9/kg bw 160 mg metformin/kg bw	Naringenin 9 ↓ FBG and improved insulin resistance. Effect comparable to metformin.	(Ren et al. 2016)
Pancreatic endoderm transplanted pregnant db/+ mice with diabetes	Control Vs. naringenin 9	–	20 days, oral	50 mg/kg bw	Naringenin 9 ↑ serum insulin levels and ↓serum glucose level in pregnant diabetic mice.	(Xing, Yang, and Wu 2016)
STZ/NA-induced diabetic rats	Control Vs. naringenin 9 Vs. glyclazide	–	21 days, oral	50 mg naringenin 9/kg bw 5 mg glyclazide/kg bw	Naringenin 9 ↓ FBG and HbA1c levels and ↑ serum insulin and antioxidant enzyme activities. Effect comparable to glyclazide.	(Annadurai et al. 2012)
LDL receptor-null (Ldlr^-/-) mice with insulin resistance	Control Vs. naringenin 9	High-fat diet	4 weeks, oral	1% and 3% wt/wt naringenin 9 in the diet	Naringenin 9 (3%) ↓insulin resistance and improved glucose tolerance.	(Mulvihill et al. 2009)
LDL receptor-null (Ldlr^-/-) mice with insulin resistance	Control Vs. nobiletin 13	High-fat diet	8 weeks, oral	0.1% and 0.3% wt/wt nobiletin 13 in the diet	Nobiletin 13 (0.3%) ↓ FBG and FBI and ↑ hepatic and peripheral insulin sensitivity.	(Mulvihill et al. 2011)
Obese diabetic ob/ob mice	Control Vs. nobiletin 13	Glucose (2 g/kg bw)	5 weeks, oral gavage	200 mg nobiletin 13/kg bw	Nobiletin 13 ↓ fasting and non-fasting plasma glucose level, HOMA-IR, and ↑ plasma adiponectin levels. ↓ mRNA of IL-6 and MCP-1 and ↑ adiponectin and PPAR-γ in adipose tissues. ↑ GLUT4 and GLUT1 in muscle and adipose.	(Lee et al. 2010)
High-fat diet-fed STZ induced type 2 diabetic rat	Control Vs. naringenin 9	Maltose or sucrose (2 g/kg bw)	2 hr, oral	25 mg naringenin 9/kg bw	Naringenin 9 reduced the total glycemic response (AUC) for both maltose and sucrose by 49%.	(Priscilla et al. 2014)
C57BL/6J mice	Control Vs. naringenin 9	High-fat diet	16 weeks, oral	1% naringenin 9 in the diet	Naringenin 9 ↓ FBG, TNF-α, MCP-1, and TLR2 in adipose tissue.	(Yoshida et al. 2013)
Wistar rats	Control Vs. grapefruit juice	3 g/kg bw of glucose	14 days, oral	3 mL/kg bw grapefruit juice	Grapefruit juice ↓ FBS, and improved glucose tolerance, with no change in insulin. Grapefruit juice did not affect intestinal glucose absorption.	(Owira and Ojewole 2009)
STZ/NA-induced type 2 diabetic rats	Control Vs. naringenin 9 Vs. glibenclamide	2 g/kg of glucose or sucrose	5 days, intragastric	50 mg naringenin 9/kg bw 5 mg glibenclamide /kg bw	Acute and chronic administration of naringenin 9 ↓ glucose intolerance compared to control.	(Ortiz-Andrade et al. 2008)
Pregnant db/+mice with gestational diabetes	Control Vs. naringenin 9	–	7 days, intraperitoneal	50 mg naringenin 9/kg bw	Naringenin 9 ↓ FBG by 15% and improved glucose intolerance by 12.5% and reduced GDM associated inflammation.	(Nguyen‐Ngo, Willcox, and Lappas 2019)
Pregnant C57BLKsJ db/+ mice with gestational diabetes	Control Vs. naringenin 9	Glucose (2 g/kg bw) and insulin (1.0 mU/kg bw)	4 weeks, oral gavage	100 mg naringenin 9/kg bw	Naringenin 9 ↓ FBS, IL-1β, IL-6, TNF-α, and MCP-1, and improved glucose and insulin tolerance.	(Li et al. 2019)
Wistar rats with cadmium-induced pancreatitis	Control Vs. hesperidin 1	–	4 weeks, oral	50 and 100 mg hesperidin/kg bw	Both hesperidin 1 doses ↑ serum insulin and pancreatic antioxidant status and reduced FBG, serum α-amylase, lipase, and inflammatory markers, IL-6, TNF-α, NF-κB, and iNOS.	(Aja et al. 2020)

Abbreviations: ↓ – Decreased; ↑ – Increased; bw – Body weight; DPP-4 – Dipeptidyl peptidase-4; F16BP – Fructose 1,6-bisphosphatase; FBG – Fasting blood glucose; FBI – Fasting blood insulin; GDM – Gestational diabetes mellitus; GK – Glucokinase; GLUT1 – Glucose transporter 1; GLUT2 – Glucose transporter 2; GLUT4 – Glucose transporter 4; HbA1c – Glycosylated hemoglobin; HMG-CoA – 3-hydroxy-3-methylglutaryl coenzyme A; HOMA-IR – Homeostatic model assessment of insulin resistance; IFN-γ – Interferon-gamma; IL-10 – Interleukin 10; IL-1β – Interleukin 1 beta; IL-6 – Interleukin 6; iNOS – Inducible nitric oxide synthase; LDL – Low-density lipoprotein; MCP-1 – Monocyte chemoattractant protein; NF-κB – Nuclear factor Kappa B; PEPCK – Phosphoenolpyruvate carboxykinase; PMF – Polymethoxyflavone; PPARα – Peroxisome proliferator-activated receptor alpha; PPARγ – Peroxisome proliferator-activated receptor gamma; STZ – Streptozotocin; STZ/NA – Streptozotocin-nicotinamide; TC – Total cholesterol; TG – Triglyceride; TLR2 – Toll-like receptor 2; TNF-α – Tumor necrosis factor-α.

Hesperidin 1 lowered glucose absorption in differentiated Caco-2/TC7 cells by inhibiting glucose transporters (Kerimi, Gauer, et al. 2019). Neohesperidin 4 inhibited absorption of D-[6-³H] glucose in Caco-2/TC7 cells by approximately 50% at a concentration of 100 µM, mainly by interacting with SGLT1. In comparison to hesperidin 1, the aglycone hesperetin 8 more potently inhibited glucose transport in Xenopus oocytes by inhibiting GLUT2, whereas fructose transport (inhibition of GLUT5) was better inhibited by hesperidin 1 (Kerimi, Gauer, et al. 2019). Similarly, hesperetin 8 strongly inhibited glucose and fructose absorption in Xenopus oocytes expressing SGLT1, GLUT2, and GLUT5 (Kwon et al. 2007). Naringenin 9 showed moderate inhibition, with no effect seen for naringin 6 (Table 2) (Kwon et al. 2007). Naringenin 9 strongly inhibited SGLT1-assisted glucose uptake in everted intestinal sleeves of normal and diabetic rats, with no effect seen with naringin 6 (J. M. Li et al. 2006).

Naringenin 9 modulated blood glucose response in normal and diabetic mice mainly by regulating intestinal glucose absorption (Ortiz-Andrade et al. 2008). Administration of naringenin 9 (0, 10, 30, and 50 mg/kg body weight (bw)) reduced intestinal glucose absorption in a dose-dependent manner (>50%) in insulin-resistant rats, tested from blood drawn from the portal vein of the rats 30 min after an oral dosage of 2 g/kg bw glucose (Kannappan and Anuradha 2010). In contrast, grapefruit juice (3 mL/kg bw), a good source of naringin 6, did not change blood glucose response in healthy rats after a glucose tolerance test (3.0 g/kg bw glucose), regardless of the route of administration (oral vs. intraperitoneal), indicating no effect of grapefruit (poly)phenols on intestinal glucose absorption (Owira and Ojewole 2009). In another study, grapefruit juice (3 mL/kg bw) improved glucose intolerance in streptozotocin (STZ)-induced diabetic rats after an intraperitoneal dose of 3.0 g/kg bw glucose, suggesting a post-absorptive mechanism from the grapefruit components in reducing glucose response (Hayanga et al. 2016). Although the aglycones (hesperetin 8 and naringenin 9) appear to show promising inhibitory effects, it should be noted that the aglycones are present in extremely low levels in the fruits naturally, and are less likely to be hydrolyzed in the small intestine to exert any effect on the starch digesting enzymes or the glucose transporters (Actis-Goretta et al. 2015; Borges et al. 2010, 2013). Another point to note is that some studies examine blood glucose in rats while anesthetized (Kannappan and Anuradha 2010). In addition to the body being under stress during anesthesia, it also reduces gastrointestinal motility and affects the digestion and absorption processes (Tweedle and Nightingale 1989). Thus, the disturbed metabolic function in rats could have also contributed to the favorable effect seen with naringenin 9. In summary, although no strong effects were seen, the few glucose transport studies done so far (Tables 2 and 3) indicate weak to moderate inhibition by the citrus components. More studies are recommended to explore the role of citrus (poly)phenols on glucose transporters using the natural juice and appropriate components of citrus fruits (not the aglycones), including the PMFs at physiologically relevant concentrations that are attainable in the gut.

Possible anti-diabetic activity of citrus (poly)phenols after absorption

In vitro and animal studies report that citrus (poly)phenols affect the markers of type 2 diabetes in different cells and tissues mainly by affecting hepatic glucose metabolism, increasing insulin secretion in the pancreas, and increasing insulin sensitivity in insulin target tissues (Tables 2 and 3). A summary of the proposed post-absorptive anti-diabetic mechanisms of citrus (poly)phenols is depicted in Figure 3. However, shortcomings in the reported studies make conclusive interpretations difficult. For example, ∼70% of the in vitro studies (post-absorptive mechanism) listed in Table 2 have utilized unrealistic (poly)phenol doses and compounds, thereby compromising physiological relevance. Thus, in the following sections, in addition to discussing the results, the study designs are critically evaluated for their physiological relevance and practical application.

Figure 3. Post-absorptive antidiabetic mechanism of citrus (poly)phenols derived from in vitro and animal models. Abbreviations: ↑ – Increase, ↓ – Decrease, PI3-K – Phosphatidylinositol 3 kinase, AMPK – Adenosine monophosphate-activated protein kinase, PPAR-γ – Peroxisome proliferator-activated receptor gamma, GLUT4 – Glucose transporter 4, GK – Glucokinase, PEPCK – Phosphoenolpyruvate carboxykinase, G6P – Glucose-6-phosphatase.

Effect on hepatic glucose homeostasis

The liver is an important metabolic organ that maintains systemic glucose homeostasis by regulating gluconeogenesis, glycogenolysis, glycolysis and other pathways (Petersen, Vatner, and Shulman 2017). The liver alone contributes to ∼90% of total endogenous glucose. Citrus flavonoids have been proposed to enhance hepatic glycolysis and glycogen content and reduce hepatic gluconeogenesis (Tables 2 and 3). Both cell and mouse experimental models suggest that citrus flavonoids modulate hepatic glucose metabolism by increasing hepatic glucokinase (GK) and reducing phosphoenolpyruvate carboxykinase (PEPCK) and glucose-6-phosphatase (G6P) mRNA expression and activity (Akiyama et al. 2009; Jung et al. 2004, 2006; do Nascimento et al. 2018; Park et al. 2013; Shen, Xu, and Lu 2012). GK, PEPCK, and G6P play a pivotal role in liver glucose homeostasis and are potential pharmacological targets for the treatment of type 2 diabetes (Rines et al. 2016; Sharabi et al. 2015). GK is involved in cellular glucose usage and PEPCK and G6P in glucose production (Sharabi et al. 2015). The diabetic state lowers GK activity, and elevates PEPCK and G6P levels (Jung et al. 2006; Rines et al. 2016). Thus, even a small change brought about by (poly)phenols in the expression of these genes could conceivably aid in improving diabetic outcomes. Citrus flavonoids, hesperidin 1, neohesperidin 4, naringin 6, and nobiletin 13 are reported to stimulate glycolysis, increase hepatic glycogen level, and reduce gluconeogenesis in HepG2 cells, mainly via modulating the activity of the enzymes, GK and G6P (Shen, Xu, and Lu 2012). The flavonoids significantly elevated the activity of GK and reduced the activity of G6P at concentrations of 10 and 50 µM, which was almost comparable to the activity of the commercial drug metformin (1 mM) (Shen, Xu, and Lu 2012), a well-known pharmacological agent used to treat type 2 diabetes acting mainly via suppressing hepatic gluconeogenesis (Petersen, Vatner, and Shulman 2017). Though the study determines a favorable effect from these compounds in hepatic glucose homeostasis, it is noteworthy that none of these compounds are found in circulation and the study dose is far above the normal range reported for the flavanone metabolites (maximum 2.2 and 6 µM for hesperetin 8 and naringenin 9 metabolites), thus making the study outcomes physiologically questionable. In general, most of the flavanone glycosides pass unaltered in the small intestine to the colon, are deglycosylated by the gut microbes, absorbed into the colonocytes, and then subjected to phase II metabolism in the colon and/or liver (Brett et al. 2008; Pereira-Caro et al. 2015, 2016; Tomás-Navarro et al. 2014; Williamson, Kay, and Crozier 2018). The main flavanone metabolites detected in plasma and urine are glucuronide, sulfate, and sulfo-glucuronide conjugated derivatives of hesperetin 8 and naringenin 9 (Figure 4).

Figure 4. Structures of main hesperetin and naringenin metabolites. Compounds detected in plasma/urine after consuming orange juice, based on data of Pereira-Caro et al. (2014, 2016).

Several studies have reported the positive effect of hesperidin 1 and naringin 6 on hepatic glucose homeostasis in normal and diabetic mice (Akiyama et al. 2009, 2010; Jung et al. 2006, 2004; Pu et al. 2012). Sweetened clarified grapefruit juice and naringin 6 showed similar anti-diabetic effects as metformin in C57BJ/6 mice fed a high-fat diet (Chudnovskiy et al. 2014). Ingestion of clarified grapefruit juice for 100 days reduced the fasting blood glucose level by 13 − 17%, fasting serum insulin level by three-fold and improved glucose intolerance, compared to the control group (Chudnovskiy et al. 2014). Hesperidin 1 and naringin 6 significantly up regulated the mRNA expression of hepatic GK in the liver of db/db mice, with naringin 6 additionally suppressing the expression of PEPCK and G6P (Jung et al. 2006). The changes observed in mRNA expression were in conjunction with the respective enzyme activities (Jung et al. 2004). Furthermore, hesperidin 1 and naringin 6 also significantly increased hepatic glycogen content compared to the diabetic control, indicating the positive effect of these two compounds on glycogenesis and glycogenolysis (Jung et al. 2004). Though naringin 6 appeared to be better in regulating hepatic glucose homeostasis than hesperidin 1, the blood glucose-lowering effect was similar for both hesperidin 1 and naringin 6 (Jung et al. 2004, 2006). Another study carried out with grapefruit juice reported improvement in fasting glucose levels and glucose tolerance, with no effect on the fasting plasma insulin levels of STZ-diabetic rats (Hayanga et al. 2016). Grapefruit juice suppressed hepatic gluconeogenesis by significantly enhancing GK activity and hepatic glycogen content and lowering G6P and PEPCK activities (Hayanga et al. 2016). Additionally, naringin 6 (0.02% in the diet) significantly reduced PEPCK and G6P mRNA expression by 47% and 62%, respectively, compared to the diabetic control mice (Pu et al. 2012).

Oral administration of 100 mg tangeretin 12/kg bw to diabetic rats for 30 days significantly decreased plasma glucose and glycosylated hemoglobin (HbA1c) levels and returned insulin level to near normal (Sundaram, Shanthi, and Sachdanandam 2014). Tangeretin 12 treatment reduced blood glucose and HbA1c level in diabetic rats by 2.4- and 2.1-fold, respectively, compared to the control. The anti-diabetic effect exerted by tangeretin 12 was comparable to the activity of the commercial drug glibenclamide (dose 5 mg/kg bw). Furthermore, tangeretin 12 treatment restored liver weight to normalcy, probably as a result of decreased glycogenolysis, protein degradation, and gluconeogenesis in the diabetic rats, and restored the activity of the hepatic carbohydrate metabolizing enzymes hexokinase, pyruvate kinase, lactate dehydrogenase, glucose-6-phosphatase, fructose-1,6-bisphosphatase, glucose-6-phosphate dehydrogenase, glycogen synthase and glycogen phosphorylase to normalcy (Sundaram, Shanthi, and Sachdanandam 2014). Similarly, the other most common PMF found in citrus fruits, nobiletin 13, was also reported to improve glucose homeostasis in obese diabetic ob/ob mice treated with nobiletin 13 (200 mg/kg bw) for 5 weeks (Lee et al. 2010). Nobiletin 13 treatment significantly reduced blood glucose concentration and HOMA-IR (homeostatic model assessment of insulin resistance), and improved glucose tolerance, plasma adiponectin levels, and also significantly decreased the mRNA expression of PEPCK and G6P in diabetic mice, with no change observed in their respective protein levels (Lee et al. 2010). The decrease in mRNA expression of PEPCK and G6P in the nobiletin 13-treated group was 66% and 71%, respectively, compared to the control. Although the PMFs appear to show promising effects in regulating hepatic glucose production and output, the doses used in both studies discussed here are quite high, not achievable just by consuming citrus fruits. The amount of PMF delivered from drinking 250 mL of orange juice is less than 2 mg (Gattuso et al. 2007). Thus, the health effects mentioned for the PMFs need to be reassessed with studies using physiologically relevant and practically achievable doses.

Similarly, some studies reporting the possible favorable effect of flavanone aglycones on liver glucose homeostasis, especially the in vitro studies, have also used extremely high concentrations, up to 300 μM (Constantin et al. 2014; do Nascimento et al. 2018). Though flavanones are abundantly present in citrus fruits, especially in grapefruit and orange fruits and juices, the peak plasma concentrations reported for naringenin and hesperetin metabolites (Figure 4, 14–26) from studies published so far range from 0.04 − 6 μM and 0.07 − 2.2 μM, respectively (Actis-Goretta et al. 2015; Erlund et al. 2001; Vallejo et al. 2010), which is far below the concentrations employed in most of the studies listed in Table 2. Even ingestion of grapefruit juice with ∼1000 μM naringenin 9, did not raise the plasma naringenin 9 level beyond 6 μM (Erlund et al. 2001).

Effect on insulin secretion

Insulin plays a key role in glucose homeostasis and is critical in the development of type 2 diabetes (Cantley and Ashcroft 2015; Fu, Gilbert, and Liu 2012; Röder et al. 2016). Insulin is the main hormone involved in lowering blood glucose levels, and is counteracted by several hormones, including glucagon, epinephrine, and glucocorticoids, which can elevate blood glucose concentration (Aronoff et al. 2004; Cantley and Ashcroft 2015). Type 2 diabetes can occur as a result of insulin insufficiency and/or impaired insulin action in target tissues such as the liver, muscle and adipose tissues. Chronic (72 hr) and acute naringenin 9 treatment, dose-dependently (10⁻¹⁰, 10⁻⁸, 10⁻⁶ M) enhanced glucose-stimulated insulin secretion in INS-1E cells (rat-derived pancreatic β-cells) (Bhattacharya et al. 2014). Naringenin 9 demonstrated a more pronounced effect on insulin secretion than caffeic acid or quercetin (Bhattacharya et al. 2014). The presence of naringenin 9 (367 μM) increased glucose-stimulated insulin secretion by ∼1.5-fold in human embryonic stem cell-derived pancreatic endoderm cells (Xing, Yang, and Wu 2016). Once again, the concentration of naringenin 9 utilized in the study is far higher than the concentration of naringenin metabolites (Figure 4, 21–26) normally found in the circulation, and so the results are unlikely to be physiologically relevant (Table 2). However, the same favorable effect was seen in the pancreatic endoderm transplanted gestational diabetic mice, where feeding 50 mg/kg bw naringenin 9 for 20 days increased the body weight and restored the blood glucose and insulin levels to normalcy (Xing, Yang, and Wu 2016).

Treatment with 0.02% hesperidin 1 and naringin 6 supplemented diet for 5 weeks increased serum insulin and C-peptide levels in the db/db mice with type 2 diabetes (Jung et al. 2004). C-peptide is widely used to assess the insulinogenic effect of compounds due to its comparatively longer half-life compared to insulin (Jones and Hattersley 2013). There was a 2-fold increase in the C-peptide level in naringenin 9-fed (100 mg/kg bw, 4 weeks) type 2 diabetic rats (Ahmed et al. 2017). Furthermore, naringenin 9 protected pancreatic islets of rats from streptozotocin-nicotinamide (STZ/NA)-induced oxidative stress and promoted regeneration of β-cells and reduced vacuolation of the pancreatic islets (Annadurai et al. 2012). Similar findings have been reported in STZ-diabetic rats fed a citrus peel extract where the extract improved insulin secretion and reversed the detrimental effect of STZ on β-cells (Sathiyabama et al. 2018). In support, several other studies have reported the positive effect of citrus extract or its (poly)phenols on the insulin level of diabetic rats (Parmar et al. 2012; Sundaram, Shanthi, and Sachdanandam 2014).

Inhibition of the dipeptidyl peptidase-4 (DPP-4) enzyme is proposed as a potential therapeutic strategy to improve blood glucose concentration (Deacon 2019), and there are several DPP-4 inhibitors (gliptins) currently in clinical use (Deacon 2019; Gupta et al. 2018). DPP-4 inactivates the incretin hormones, glucagon-like-peptide-1 (GLP-1), and glucose-dependent insulinotropic polypeptide (GIP), which are involved in increasing insulin secretion and sensitivity, and simultaneously reduces glucagon secretion from the pancreas. Citrus flavonoids potentially inhibit DPP-4 activity in vitro and in vivo (Fan et al. 2013; Gupta et al. 2018; Parmar et al. 2012; Proença et al. 2019). The citrus flavonoids, hesperetin 8, and naringenin 9 inhibited DPP-4 activity ∼15-fold more effectively than the positive control Diprotin A (Fan et al. 2013) (Table 2). Another study reported that naringin 6 inhibited DPP-4 activity to a similar extent as sitagliptin (Parmar et al. 2012). In the same work, feeding naringin 6 (40 mg/kg bw) for 9 days significantly reduced the serum DPP-4 activity by 31% and increased serum insulin level by 68% in Wistar rats compared to the control. However, recent studies have reported citrus flavonoids to be extremely weak inhibitors of human recombinant DPP-4 (Gupta et al. 2018; Proença et al. 2019). Fan et al. (2013) used DPP-4 derived from porcine kidneys, and the in vitro study by Parmar et al. (2012) does not mention the source. As discussed before, the activity of the (poly)phenols varies widely based on the source of the enzyme. Results gained from studies using human-derived enzymes are considered more applicable to human nutrition. Thus, it is difficult to comment on the effect of citrus (poly)phenols on DPP-4 based on just two studies, and further work is recommended to explore the effect of citrus (poly)phenols on DPP-4.

Effect on insulin sensitivity

Citrus (poly)phenols improve insulin sensitivity in the liver and peripheral tissues by modulating several metabolic pathways through the expression of various genes and proteins (Ahmed et al. 2017; Erlund et al. 2001; Jung et al. 2004, 2006; Shen, Xu, and Lu 2012; Zygmunt et al. 2010). Citrus sinensis (L.) Osbeck (sweet orange) peel extract significantly improved HOMA-IR in STZ-diabetic rats, which was comparable to metformin (Sathiyabama et al. 2018). Further, compared to the diabetic control, citrus extract supplementation in the diet (50 and 100 mg/kg b.w) for 30 days significantly up-regulated peroxisome proliferator-activated receptor gamma (PPAR-γ), GLUT4, and insulin receptor mRNA and protein in diabetic mice (Sathiyabama et al. 2018). The glucose-lowering effect of the citrus extract was postulated to be through activation of PPAR-γ, which in turn stimulated adipocyte differentiation, insulin receptor signaling, and translocation of GLUT4 from the cytosol to the plasma membrane in the adipose tissue in an insulin-dependent manner via affecting the phosphoinositide-3-kinase/protein kinase B (PI3-k/Akt) pathway (Sathiyabama et al. 2018). PPARs are a group of nuclear hormone receptors involved in regulating the expression of genes involved in glucose homeostasis, predominantly expressed in adipose tissues but also found in other tissues such as the liver, heart, and skeletal muscle (Kim and Ahn 2004; Vidal-Puig et al. 1997). PPAR-γ improves the insulin-sensitizing effects by increasing the expression of GLUT4 in adipose tissues, directly activates hepatic GK expression, inhibits hepatic gluconeogenesis, and also increases the glucose-sensing ability of pancreatic β-cells (Jung et al. 2006; Kim and Ahn 2004). In another study, hesperidin 1 and naringin 6 were reported to act as agonists for PPAR-γ in db/db mice (Jung et al. 2006). The two (poly)phenols significantly increased PPAR-γ protein expression by about 3-fold, but not PPAR-α, in the liver and adipose tissues of diabetic mice. Hesperidin 1 and naringin 6 reduced GLUT2 protein expression in hepatocytes of diabetic mice, a transporter that is normally upregulated in hepatocytes in a diabetic state (Chadt and Al-Hasani 2020), and increased hepatic GK mRNA expression, thereby reducing hepatic glucose output (Jung et al. 2006). Moreover, GLUT4 protein expression was increased in adipocytes by hesperidin 1 and naringin 6 compared to the control group, indicating enhanced insulin sensitivity and uptake of glucose into adipose tissues in the flavanone treated group (Jung et al. 2006).

Key pathways modulated by citrus (poly)phenols affecting insulin sensitivity Citrus (poly)phenols are reported to improve insulin sensitivity by modulating the PI3-k/Akt, the protein kinase A (PKA), and the adenosine monophosphate-activated protein kinase (AMPK) signaling pathways in insulin target tissues (Kim et al. 2012; Onda et al. 2013; Zygmunt et al. 2010). The AMPK, PKA, and the PI3-k/Akt pathways are potential drug targets in the management of type 2 diabetes (Coughlan et al. 2014; Huang et al. 2018; Sayem et al. 2018; Yang and Yang 2016). AMPK is an intracellular energy-sensing enzyme, which increases glucose transport and GLUT4 translocation through an insulin-independent mechanism, while the PI3-k/Akt pathway is insulin-dependent (Coughlan et al. 2014; Sayem et al. 2018). Both AMPK and PI3-k/Akt pathway activation improves insulin sensitivity and metabolic health by stimulating intracellular glucose transport and metabolism in skeletal muscle, fatty acid oxidation in adipose tissues and reduces gluconeogenesis, and stimulates lipolysis in the liver (Coughlan et al. 2014; Huang et al. 2018; Sayem et al. 2018). Thus, the ability of naturally-occurring (poly)phenols to potentially modify these pathways beneficially can be considered an effective dietary strategy to control the development of type 2 diabetes.

Naringenin 9 stimulated glucose uptake in murine muscle cells (L6 myotubes) in a dose-dependent manner (10 − 150 μM) via activating the AMPK pathway (Zygmunt et al. 2010). Silencing of AMPK using small interference RNA (siRNA) prevented glucose uptake in the cells, further confirming the importance of the AMPK pathway in the naringenin 9-stimulated glucose uptake. At a concentration of 75 μM, naringenin 9 exhibited maximum glucose uptake, which was similar to the activity exerted by 100 nM insulin (Zygmunt et al. 2010). Several other studies have also reported the insulin-like effects of naringenin 9 in HepG2 cells (Allister et al. 2005; Borradaile, de Dreu, and Huff 2003) and rat adipocytes (Lim, Soh, and Kuppusamy 2008). In addition to its insulin-like effects, naringenin 9 significantly enhanced insulin-stimulated glucose uptake in muscle cells at low insulin levels (<10 nM) (Zygmunt et al. 2010). Likewise, naringenin 9 increased insulin-stimulated glucose uptake in TNF-α-stimulated insulin-resistant C2C12 myotubes through an AMPK-dependent mechanism (Li et al. 2019). Naringin 6 supplementation for 20 weeks at a level of 0.02% in the diet attenuated metabolic syndrome in C57BL/6J mice fed a high-fat diet by activating the AMPK pathway, and by inhibiting the mitogen-activated protein kinase (MAPK)-dependent signaling cascades, including ERK1/2, JNK1/2 and P38 pathways (Pu et al. 2012).

In differentiated C2C12 myotubes, a very high concentration of tangeretin 12 (>100 µM) stimulated AMPK activation, promoted GLUT4 translocation from cytosol to the cell membrane and glucose uptake without affecting PPAR-γ expression (Kim et al. 2012). In addition, tangeretin 12 (200 mg/kg diet) improved insulin resistance, reduced fat accumulation, and adipocytokine production, especially, leptin and resistin, and reduced the levels of the inflammatory markers MCP-1 and IL-6 in obese mice (Kim et al. 2012). Similarly, both tangeretin 12 and nobiletin 13 increased adiponectin secretion and decreased MCP-1 levels in 3T3-L1 adipocytes, whilst nobiletin 13 at a concentration of 128 µM reduced the level of resistin (Miyata et al. 2011). Adipocyte-derived cytokines, such as leptin and resistin, are important markers for the development of type 2 diabetes (Kocot et al. 2017). Resistin increases insulin resistance in peripheral tissues by affecting AMPK activity and also reduces insulin receptor substrate (IRS-1 and IRS-2) expression, thereby disrupting insulin signaling, leading to increased hepatic gluconeogenesis and reduced glycolysis. In 3T3-F442A adipocytes, tangeretin 12 and nobiletin 13 significantly enhanced glucose uptake from a concentration starting from 5 µM, mainly via activating the PI3-k/Akt and PKA pathways without affecting AMPK (Onda et al. 2013). According to the evaluated studies, high concentrations of citrus (poly)phenols appear to modulate glucose homeostasis mainly by affecting the AMPK pathway in muscle tissues, and the PI3-k/Akt pathway in adipose tissues (Table 2). However, the reported active concentrations are far above the physiological level.

Insulin sensitizing effect in adipocytes

Adipose tissue stores fat and is an endocrine organ that plays an important role in obesity-associated insulin resistance (Okabe et al. 2014). Mature healthy adipose tissues improve insulin resistance by secreting insulin-sensitizing factors, such as adiponectin and leptin. On the other hand, hypertrophied adipose tissues infiltrated by macrophages mainly secrete the insulin resistance factors, TNF-α, retinol-binding protein 4 (RBP4), IL-6, and IL-1β (Hardy, Czech, and Corvera 2012; Smith and Kahn 2016). Therefore, modulation of adipogenesis and adipocyte hypertrophy can be regarded as useful strategies in the prevention and/or control of obesity-associated insulin resistance. Citrus flavanones are reported to affect adipogenesis and adipocyte gene expression in vitro and in vivo (Ahmed et al. 2017; Akiyama et al. 2010; Liu et al. 2008; Vidal-Puig et al. 1997). Navel orange peel extract, naringin 6, and naringenin 9 were reported to significantly increase adiponectin, insulin receptor and GLUT4 mRNA expression in STZ/NA-induced type 2 diabetic rats compared to the diabetic control (Ahmed et al. 2017). Adiponectin secreted by adipocytes improves glucose and lipid homeostasis through its insulin-sensitizing properties. A lower level of adiponectin is directly associated with the development of type 2 diabetes and metabolic syndrome (Frankenberg, Reis, and Gerchman 2017; Liu et al. 2008). Hesperidin 1 and its aglycone hesperetin 8 significantly increased serum adiponectin levels to normalcy, comparable to the non-diabetic control (Akiyama et al. 2009). Liu et al. (2008) reported significant up regulation of adiponectin expression in response to 80 µM hesperetin 8 and naringenin 9 in 3T3-L1 adipocytes, comparable to the commercial drug rosiglitazone, but this concentration is an order of magnitude above that achievable physiologically. Nobiletin 13 improved hyperglycemia and insulin resistance in diabetic ob/ob mice by increasing plasma and adipose adiponectin levels, and by up regulating the expression of GLUT1 and GLUT4 in both muscle and adipose tissues (Lee et al. 2010). Nobiletin 13 enhanced differentiation of ST-13 preadipocytes into mature adipocytes and increased the adiponectin level in 10 μM nobiletin 13-treated cells by 3.4-fold compared to the control (Kunimasa et al. 2009).

In contrast to reports on the beneficial effects of citrus (poly)phenols, several in vitro cell-based studies also report anti-insulin sensitizing effects using MDA-MB-231 breast cancer cells (Yang et al. 2013), MC3T3-G2/PA6 adipocytes (Nomura et al. 2008), U937 cells (Park 1999) and 3T3-L1 adipocytes (Claussnitzer et al. 2011; Harmon and Patel 2003; Kanda et al. 2012; Liu et al. 2008; Richard et al. 2013). A significant dose-dependent reduction in insulin-stimulated glucose uptake and IRS-1 tyrosine phosphorylation was reported in mature 3T3-L1 adipocytes in response to a high concentration of naringenin 9 (44, 92 and 183 µM) (Richard et al. 2013). Further, high concentrations of naringenin 9 also reduced adiponectin secretion in both human (92 µM naringenin 9) and murine adipocytes (183 µM naringenin 9) (Richard et al. 2013). Similarly, 20 µM naringenin 9 was found to significantly reduce basal and insulin-stimulated glucose uptake (∼20% inhibition) in both 3T3-L1 and mature human adipocytes (Claussnitzer et al. 2011). The inhibitory effect exerted by naringenin 9 was even prominent at a concentration as low as 1 µM (Claussnitzer et al. 2011). Several other studies have reported similar effects with naringenin 9 (Table 2) and the anti-insulin sensitizing effect exhibited by naringenin 9 in adipocytes has been postulated to be due to impaired PI3K/Akt pathway and reduced trafficking of GLUT4 to the plasma membrane (Harmon and Patel 2003; Nomura et al. 2008; Richard et al. 2013). Claussnitzer et al. (2011) also proposed a direct inhibitory effect of naringenin 9 on GLUT4. Likewise, hesperetin 8 was reported to downregulate the expression of GLUT1, inhibit insulin-induced translocation of GLUT4, and suppress phosphorylation of the insulin receptor beta subunit and Akt in MDA-MB-231 breast cancer cells (Yang et al. 2013).

In contrast to the reports on the lowering of insulin sensitivity by naringenin 9 on adipose tissue, Liu et al. (2008) reported that both hesperetin 8 and naringenin 9 significantly increased adiponectin mRNA expression and increased PPAR-γ-controlled luciferase expression in a dose-dependent manner in 3T3-L1 adipocytes. Naringenin chalcone, an intermediate derivative of naringenin 9, was reported to enhance insulin-sensitizing effects in 3T3-L1 adipocytes by increasing the expression of adiponectin, both at a transcriptional and translational level (8- and 2.2-fold, respectively), without affecting preadipocyte differentiation up to a concentration of 100 µM (Horiba et al. 2010). Microarray studies revealed further up-regulation of several genes linked with energy metabolism (Horiba et al. 2010). At a concentration of 100 µM, naringenin 9 treatment led to increased glucose uptake (163%) in primary rat adipocytes, superior to insulin (Lim, Soh, and Kuppusamy 2008). Furthermore, naringenin 9 significantly increased glucose uptake in a dose-dependent manner (0.01 − 100 µM naringenin 9) independent of insulin (Lim, Soh, and Kuppusamy 2008). Naringenin 9 at >10 µM reduced Toll-like receptor 2 (TLR2) mRNA and protein expression in 3T3-L1 adipocyte during differentiation (Yoshida et al. 2013). Toll-like receptors are directly linked with the obesity-associated inflammatory response in adipose tissues, which long term can lead to insulin resistance and type 2 diabetes. The authors saw the same effect being translated in high-fat diet-fed C57BL/6J mice, where supplementation of the diet with 1% naringenin 9 for 16 weeks improved glycemic response and reduced the level of TNF-α, MCP-1, and TLR2 in the adipose tissues of obese mice (Yoshida et al. 2013).

In rodents, many studies report a favorable glycemic outcome upon long-term supplementation with citrus extract or pure compound, including naringenin 9 (Table 3). Naringenin 9 supplementation for 44 days at 0.05% of the diet increased insulin signaling and sensitivity in fructose-fed insulin-resistant rats (Kannappan and Anuradha 2010). Other studies with STZ-treated (Ren et al. 2016), STZ-nicotinamide treated (Annadurai et al. 2012), diabetic pregnant db/+ (Nguyen‐Ngo, Willcox, and Lappas 2019; Xing, Yang, and Wu 2016), OFA (Mayneris-Perxachs et al. 2019) and LDL-receptor null mice (Mulvihill et al. 2009) also demonstrated the positive effect of naringenin 9 on reducing the incidence of insulin resistance and type 2 diabetes (Table 3). Depending on the molecular structure and chemical properties, (poly)phenols are capable of modulating the activity or expression of various genes/proteins, and the outcome will depend on the cumulative effect. When the beneficial effects of a compound or a group of compounds outweigh any opposing effects, improvements in the biomarkers of interest are seen. Thus, although cell culture studies sometimes show a negative effect of naringenin 9 on insulin resistance, as a whole, citrus extract and its compounds seem to improve glucose homeostasis, especially in vivo.

Study limitations

The most important point to note is that the concentrations of the (poly)phenols used in most of the in vitro studies reported here are extremely high. Out of the 33 studies listed in Table 2 on the potential post-absorptive mechanisms, 23 studies have employed physiologically unrealistic (poly)phenol treatments to show an effect. In general, (poly)phenols or their metabolites reach the circulation or target tissues in low concentrations, certainly considerably below 10 µM. According to bioavailability data, the maximum plasma concentration attained for hesperetin 8 and naringenin 9 metabolites range from 0.07 − 2.2 µM and 0.04 − 6 µM (Actis-Goretta et al. 2015; Erlund et al. 2001; Vallejo et al. 2010), respectively, while in vitro studies have used concentrations ranging from 1 µM to >300 µM, with significant effects generally only seen at concentrations toward the higher end (Table 2). Since the amount of (poly)phenols found within the gut is generally at least several-fold higher than present in the circulation, the use of higher concentrations in studies looking at effects within the gut is acceptable to a certain extent. Generally, 300 mL of orange juice delivers >300 µmol and >50 µmol of hesperidin 1 and naringenin 9, respectively (Aschoff et al. 2015; Gil-Izquierdo et al. 2001). Assuming the bioaccessibility of these compounds to be 100%, they will be found in high concentrations in the lumen. On the other hand, studies done on non-gut cells/enzymes which focus on post-absorptive effects should utilize physiologically applicable concentrations, usually of the same order as the maximum plasma concentration (C_max). Unfortunately, the majority of the studies on non-gut cells have employed non-physiological doses of (poly)phenols (Table 2), bringing into question the physiological and clinical relevance in humans.

Further, some studies have also utilized flavanone glycosides in non-gut cells. Use of the aglycones is acceptable to a certain extent, but not the glycosides. To obtain meaningful results, instead of the aglycones, physiologically relevant flavanone conjugates (14–26) and their microbial catabolites should be utilized, as the bioactivity exerted by the metabolites could be different from the parent compound and/or the aglycones. Thus, to identify the potential effects and mechanisms, future studies should emphasize employing physiologically relevant concentrations (<10 µM) of the flavanone metabolites/catabolites.

Overall summary and future recommendations

This paper critically reviews studies reporting the mechanism of effects of citrus (poly)phenols on modulating the risk of type 2 diabetes. Over the past two decades, many studies have focused on identifying the effect of citrus (poly)phenols, especially hesperidin 1/hesperetin 8, narirutin 5/naringin 6/naringenin 9, and the polymethoxyflavones, tangeretin 12, and nobiletin 13 on markers of type 2 diabetes risk and their potential mechanisms of action. Despite numerous studies, shortcomings in study designs make conclusive interpretations difficult. With regard to in vitro studies, although they indicate a possible beneficial effect through identified and proposed mechanisms, about 70% of the studies listed in Table 2 have utilized physiologically unrealistic concentrations and/or forms of the compounds. Except for studies focusing on the mechanisms within the gut (starch hydrolyzing enzymes and glucose transporters), studies conducted with non-gut cells and enzymes have utilized very high doses, and the concentrations required to see an effect are far above the physiological concentrations. Another point to note is the inappropriate use of compounds in the studies, without considering physiological relevance; there is no point in treating liver cells with hesperidin 1 for example since after consumption, hesperidin 1 is deglycosylated in the gut before absorption. This means that hesperetin 8 and conjugates are found in the blood, and are the chemical species that will encounter liver cells, but not the parent hesperidin. Thus, studies targeting the intestinal lumen should focus on using the flavanone glycosides instead of the aglycones, as these compounds are less likely to be hydrolyzed in the small intestine, and studies focused on post-absorptive mechanisms should utilize the aglycones or their respective metabolites. According to bioavailability studies, the majority of the citrus flavanones undergo Phase II metabolism and the methylated, sulfated, or glucuronidated metabolites are the primary compounds found in the circulation. Going forward, studies should focus more on using the flavanone metabolites and especially the microbiota-derived phenolic catabolites over the aglycones, as the latter is usually at low concentration in the circulation. Hence, future in vitro studies probing the antidiabetic mechanism of citrus (poly)phenols should pay more attention to using physiologically relevant doses and focus on using the appropriate compounds or their metabolites in the experiments, as the bioactivity and efficacy of these compounds depend on their chemical and structural nature.

Overall, the in vitro and animal studies demonstrate a potential beneficial effect of citrus (poly)phenols on the markers of type 2 diabetes, predominantly through post-absorptive mechanisms, mainly by regulating hepatic glucose homeostasis and insulin sensitivity. Within the gut, the flavanones appear to moderately inhibit the glucose transporters, with little effect on starch digesting enzymes. Furthermore, based on the outcomes from animal studies, the beneficial effects derived from consuming citrus compounds appear to be more evident in long-term studies as compared to acute treatments.

Abbreviations
ADP	=	Adenosine diphosphate
Akt	=	Protein kinase B
AMPK	=	5’ adenosine monophosphate-activated protein kinase
apoB	=	Apolipoprotein B
bw	=	Body weight
CD	=	Clathrate
CGO	=	Citrus grandis Osbeck
ChREBP	=	Carbohydrate response element-binding protein
Cmax	=	Maximum plasma concentration
DMSO	=	Dimethyl sulfoxide
DPP-4	=	Dipeptidyl-peptidase 4
EGCG	=	Epigallocatechin gallate
ERK1/2	=	Extracellular signal-regulated protein kinase
F16BP	=	Fructose-1,6-bisphosphatase
FBG	=	Fasting blood glucose
FBI	=	Fasting blood insulin
G6P	=	Glucose-6-phosphatase
GDM	=	Gestational diabetes mellitus
GIP	=	Glucose-dependent insulinotropic polypeptide
GK	=	Glucokinase
GLP-1	=	Glucagon-like-peptide-1
GLUT1	=	Glucose transporter 1
GLUT2	=	Glucose transporter 2
GLUT4	=	Glucose transporter 4
GLUT5	=	Glucose transporter 5
GSK-3	=	Glycogen synthase kinase-3
HbA1c	=	Glycosylated hemoglobin
HMG-CoA	=	3-hydroxy-3-methylglutaryl coenzyme A
HOMA-IR	=	Homeostatic model assessment of insulin resistance
IFN-γ	=	Interferon-gamma
IL-10	=	Interleukin 10
IL-1β	=	Interleukin 1 beta
IL-6	=	Interleukin 6
iNOS	=	Inducible nitric oxide synthase
Ins1/2	=	Insulin
IRS-1/2	=	Insulin receptor substrate
JNK1/2	=	c-Jun N-terminal kinase
LDL	=	Low-density lipoprotein
MAPK	=	Mitogen-activated protein kinase
MCP-1	=	Monocyte chemoattractant protein
miR-140	=	MicroRNA-140
mRNA	=	Messenger ribonucleic acid
MTP	=	Microsomal triglyceride transfer protein
NF-κB	=	Nuclear factor Kappa B
PEPCK	=	Phosphoenolpyruvate carboxykinase
PI3-K	=	Phosphatidylinositol 3 kinase
PKA	=	Protein kinase A
PMF	=	Polymethoxyflavone
PPARα	=	Peroxisome proliferator-activated receptor alpha
PPAR-γ	=	Peroxisome proliferator-activated receptor gamma
PTP1B	=	Protein tyrosine phosphatase 1B
RBP4	=	Retinol-binding protein 4
SGLT1	=	Sodium-dependent glucose cotransporter
siRNA	=	Small interference RNA
SREBP-1	=	Sterol regulatory element-binding protein
STZ	=	Streptozotocin
STZ/NA	=	Streptozotocin-nicotinamide
TC	=	Total cholesterol
TG	=	Triglyceride
TLR-2	=	Toll-like receptor 2
TNF-α	=	Tumor necrosis factor-alpha.

Acknowledgements

We thank Monash University for an International PhD Scholarship to RV.

Disclosure statement

The authors declare no conflicts of interest.