*2.5. CYP450s*

Growing evidence has demonstrated that diabetes increases the expression of some CYP450s (such as CYP2A1, CYP1A2, CYP2B1/2, CYP2C6, CYP2C7, CYP3A1/2 and CYP2E1) in rats [9,11,12,26–30], although opposite reports have also been shown [31,32]. The induction of CYP3A1/2 led to lower plasma exposure of simvastatin [12], atorvastatin [9,11], verapamil [27,29] and lidocaine [30] following intravenous dose to diabetic rats. The formation of 1,3-dimethyluric acid from theophylline is mainly mediated via CYP1A2 and CYP2E1. Both alloxan-induced diabetes and STZ-induced diabetes were reported to increase expressions of hepatic CYP2E1 and CYP3A2 by three-fold, leading to a lower area under the curve (AUC) of theophylline and a higher AUC of 1,3-dimethyluric acid following oral and intravenous administration to rats [28]. The induction of CYP1A2 (and CYP3A1/2) may contribute to a lower AUC of oltipraz [33], higher non-renal clearances of omeprazole [34] and systemic clearance of antipyrine [35] in diabetic rats. Diazinon is metabolized to toxic metabolite diethylphosphate via CYP3A2 and CYP1A2. The increases in diazinon toxicity and urinary recovery of its metabolite diethylphosphate in diabetic rats may be attributed to the enhancement of hepatic CYP1A2 (and CYP3A2)-mediated metabolism of diazinon [35]. A report showed that protein expression and activity of hepatic CYP2E1 in STZ-induced diabetic rats were increased by three- and 2.5-fold in comparison to control rats, respectively. The increases were in line with the increased oxidative stress, demonstrating crucial roles of CYP2E1 in stress-induced pathological processes under diabetic status [31,32,36].

Hyperketonemia (such as acetone) is considered as a reason for diabetes inducing CYP2E1. Data from primary cultured rat hepatocytes demonstrated that insulin decreased the expression of CYP2E1 and CYP2B protein and mRNA, inferring that a deficiency of insulin is also a reason that diabetes induces the expression of CYP2El and CYP2B [37]. Furthermore, data from both Fa2N-4 cells and HepG2 cells demonstrated that fatty acids (palmitic acid, oleic acid, stearic acid and linoleic acid), but not insulin, upregulated the activity and expression of CYP3A4 mRNA and protein, indicating that increased levels of fatty acids may be one of the reasons that diabetes elevated the function and expression of CYP3A4 [38]. Recent studies showed that diabetes reduced the expression of hepatic peroxisome proliferator activated receptor γ (PPARγ) protein [39,40] and PPARα protein [41]. The decreased expression of hepatic PPARα was attributed to the increased palmitic acid [41]. A report showed that PPARα agonist gemfibrozil in PPARα-null mice showed more induction of the mRNA, protein and activity of CYP3a, CYP2b and CYP2c than in wild-type mice, indicating that the induction of gemfibrozil on CYP450s was suppressed by PPARα activation [42]. The downregulation of PPARα protein expression by diabetes [41] might attenuate the inhibition of PPARα on the expression of CYP450s, leading to the induction of hepatic CYP450s, which needs further investigation. Moreover, diabetes was reported to increase the expression of hepatic glucocorticoid receptor and increase the circulating level of corticosterone [43,44]. The increased level of corticosterone might induce expressions of CYP2B [45] and CYP2C [46] via activating glucocorticoid receptor. All these results demonstrate that alterations in the expressions of hepatic CYP450s were involved in various mechanisms. In contrast to CYP2C6 induction, diabetes significantly lowered the expression of CYP2C11 [26,47–50], impairing the metabolism of diclofenac [49], glibenclamide [48] and nateglinide [47]. The increase in CYP2C6 expression and decrease in CYP2C11 expression may partly explain why the AUC values of both phenytoin and 4′ -hydroxylphenytoin in diabetic rats were comparable to control rats [51].

Some contradictory results have also been reported. In Goto-Kakizaki rats, it was found that expressions of hepatic CYP reductase and CYP3A2 were significantly upregulated, accompanied by increases in the activities of midazolam 4-hydroxylase and CYP reductase. In contrast, hepatic expressions of CYP1A2 and CYP3A1 were downregulated, and the activities of 7-methoxyresorufin-*O*-demethylase and 7-ethoxyresorufin-O-deethylase were decreased. Expressions of other CYP450s, such as CYP1B1, CYP2B1, CYP2C11 and CYP2E1, were unaltered, inferring that diabetes regulates the expression of hepatic CYP450s in an isoform-specific manner [30,52]. Similarly, in 20-week-old male db/db mice, testosterone-6β-hydroxylation but not midazolam-1hydroxylation was significantly decreased, the intrinsic clearance (CLint) value of testosterone-6βhydroxylation (CYP3A11) was less than 46% of control mice. Other metabolic reactions such as dextromethorphan-*O*-demethylation (CYP2D), phenacetin-*O*-deethylation (CYP1A2), coumarin-7 hydroxylation (CYP2A4/5), tolbutamide-4-hydroxylation (CYP2C), bupropion-hydroxylation (CYP2B10) and omeprazole-5-hydroxylation (CYP2C29), chlorzoxazone-6-hydroxylation (CYP2E1) were unaltered [53]. Expressions and functions of hepatic CYP450s are dependent on diabetic progression. It was reported that 25-week-old db/db mice showed substantial decreases in mRNA expressions and functions of CYP2B10 and CYP2C29 compared with those of the 10-week-old db/db mice. But mRNA expressions of CYP3A11 and CYP2E1 in the 25-week-old db/db mice were comparable to those of the 10-week-old db/db mice [54]. In Zucker diabetic fatty (ZDF) rats, CYP1A1/2, and CYP3A1 levels were similar among the livers of 5 and 11-week-old ZDF rats and control rats, although 11-week-old ZDF rats possessed the higher serum glucose and glycated hemoglobin levels, but not 5-week-old ZDF rats. Moreover, 5-week-old ZDF rats, but not 11-week-old ZDF rats, showed lower levels of cytochrome b5, CYP2B1, CYP2C11, CYP2E1 and CYP3A2. Consistently, pentoxyresorufin O-depentylation, testosterone 2α- and 16α-hydroxylation, chlorzoxazone 6-hydroxylation, and midazolam 1'- and 4-hydroxylation were decreased only in 5-week-old ZDF rats [55]. Expressions (mRNA and proteins) of CYP3A11/13 and their activities were significantly increased in STZ-induced diabetic C57BLKS/J mice and db/db mice [56]. In contrast, Wang et al. reported that the levels of hepatic CYP3A11 mRNA in C57BL/6 mice diabetic mice induced by HFD/STZ were less than 30% of control mice [31]. All these results indicate that alterations in the expressions of hepatic CYP450s are highly dependent on the type of diabetes, diabetic progression and animal species.

Expressions of hepatic CYP450s in diabetes patients have been widely investigated, although some results are often in contrast to findings in diabetic mice. Gravel et al. [57] assessed the activities of seven major CYP450s (CYP1A2, CYP2B6, CYP2C9, CYP2C19, CYP2D6, CYP2E1, and CYP3A4/5) in 38 type 2 diabetic patients and 35 non-diabetic subjects following the oral administration of a cocktail of probes (caffeine, bupropion, tolbutamide, omeprazole, dextromethorphan, chlorzoxazone and midazolam). They found that the activities of CYP2C19, CYP2B6, and CYP3A in diabetic patients were decreased by about 46%, 45%, and 38%, respectively. CYP1A2 and CYP2C9 activities showed a trend to slightly increase in subjects with diabetes, but activities of CYP2D6 and CYP2E1 were unaltered [57]. Meanwhile, obese children showed higher oral clearance of chlorzoxazone and lower AUC of chlorzoxazone. The AUC was only 46% of nonobese peers, inferring the increased CYP2E1 activity [58].

In vitro data demonstrated that protein expressions and activities of CYP3A4 in hepatic microsomes of diabetic livers were significantly lower than those in non-diabetic donors, although a great variability (8.2-fold) in CYP3A4 protein levels was observed. However, the expression and activity of CYP2E1 were significantly increased [59]. The disposition of nisoldipine in humans was investigated. The results show that the disposition of nisoldipine is enantioselective and nisoldipine is preferentially metabolized. Compared with non-diabetes patients, AUC values of (+)-nisoldipine and (−)-nisoldipine in diabetes patients were increased by 94% and 143%, respectively. Similar increases were also found in their peak concentration (Cmax). Lidocaine tests demonstrated a decrease in activity of CYP3A (and CYP1A2) [60]. Similarly, pregnant women with gestational diabetes showed significantly higher Cmax and AUC of lidocaine following peridural anesthesia administration of lidocaine than non-diabetic patients [61]. However, oral plasma concentrations of nifedipine [62] and metoprolol [63] in pregnant women with gestational diabetes were comparable to those in non-diabetic pregnant women. Adith et al. [64] investigated phenytoin kinetics in 10 male type 1 and 10 type 2 diabetic patients. Age- and sex-matched epileptic patients receiving phenytoin alone served as control groups. They found that steady-state concentrations of phenytoin were significantly lower in both types of diabetic patients compared to the respective controls. In line with this, diabetes patients showed significantly increased Vmax/K<sup>m</sup> values of phenytoin. Similarly, the AUCs of cyclosporin A in kidney transplant recipients with diabetes

mellitus were approximately 50%~60% that of non-diabetic patients [65,66], which was partly attributed to the delayed gastric emptying [66]. Moreover, no difference in cyclosporine A trough levels between renal transplant patients and renal transplant patients with diabetes was found [67]. Diabetes patients were also reported to show a trend to increase in the expression and activity of CYP3A4, although no statistical significance was observed. Importantly, significant decreases were found in diabetic patients with non-alcoholic fatty liver or non-alcoholic steatohepatitis. Furthermore, the decrease in activity and protein level of CYP3A4 continued with the severity of disease as it progressed from non-alcoholic fatty liver to non-alcoholic steatohepatitis [68]. Matzke et al. [69] also investigated the activities of CYP1A2 and CYP2D6 in type 1 diabetes and type 2 diabetes patients using the probes antipyrine, caffeine and dextromethorphan. The results show that compared with controls, oral clearance of antipyrine was increased 72% in type 1 diabetes patients. Formation clearances of 4-hydroxyantipyrine and 3-hydroxymethylantipyrine were increased by 74% and 137%, respectively. The caffeine metabolic index (paraxanthine/caffeine) was increased by 34%, but no statistical significance was obtained. These alterations did not occur in type 2 diabetes patients. Similarly, although plasma clearance of theophylline in diabetic patients was comparable to that in healthy subjects, there was a positive correlation between hemoglobin A1c values and formation clearances of both 1,3-dimethyluric acid and 1-methyluric acid [70], in line with inductions of CYP1A2 and CYP2E1.

Although antipyrine is widely used to assess the activity of CYP450s in diabetic patients, the results are often confusing [69,71–73]. Sotaniemi et al. [73] assessed the effects of diabetes on hepatic drug metabolism clearance in men using antipyrine in 298 diabetic patients, who were classified by type of the disease, age, gender, duration of therapy and liver involvement. They found that a 13-fold individual variation in antipyrine metabolism existed among all the diabetic patients. Antipyrine was eliminated faster in untreated type 1 patients, which was reversed by insulin treatment. Males aged 16–59 years, who responded insufficiently to insulin therapy, had a rapid antipyrine elimination, which could be normalized by the readjustment of insulin. The antipyrine elimination rate in women with insufficient glucose control on insulin therapy was comparable to controls. In type 2 diabetic patients, the clearance of antipyrine was unaltered in women, but men over 40 years of age showed a reduced antipyrine metabolism. Diet/drinking-habits also affect the expression of hepatic CYP450s. Urry et al. [74] investigated CYP1A2 activity in diabetic patients using coffee as a probe in 57 type 2 diabetes and 146 non-type 2 diabetes. They found that diabetes patients showed higher activity of CYP1A2 than control groups. Further studies showed that participants habitually consuming more caffeine showed higher CYP1A2 activity than participants with lower caffeine consumption. Several studies have demonstrated that CYP2D6 activity is unaltered by diabetes [43,69,75].

#### *2.6. Interplay of Transporter-CYP450s in Liver*

In the liver, drugs enter hepatocytes from portal blood through passive diffusion or transporter-mediated transport. In hepatocytes, drugs are metabolized by metabolic enzymes. Parent drugs and their metabolites are pumped out of hepatocytes to bile through efflux transporters (such as P-gp, BCRP or MRP2) and are exported to the blood through passive diffusion or efflux transporters (such as MRP3 and MRP4) (Figure 1B). Typical examples are statins. Most statins are substrates of CYP450s, MRPs, OATPs, BCRP and P-gp, characterizing transporter–enzyme interplay. The overall hepatic intrinsic clearance (CLint,all) is a hybrid parameter which is composed of intrinsic influx clearances (CLint,up) from portal blood, intrinsic metabolism clearance (CLint,met), biliary excretion clearance (CLint,bile) and intrinsic efflux clearance (CLint,back) of unbound drugs from hepatocytes to the blood, i.e.,

$$\text{CL}\_{\text{int},\mu\text{l}} = \text{CL}\_{\text{int},\mu p} \times \frac{\text{CL}\_{\text{int},\text{bile}} + \text{CL}\_{\text{int},\text{met}}}{\text{CL}\_{\text{int},\text{bile}} + \text{CL}\_{\text{int},\text{met}} + \text{CL}\_{\text{int},\text{buck}}} \tag{1}$$

According to the relative values of CLint,bile and CLint,met*,* to CLint,back, the drugs are classified into three groups.


Diabetes significantly disorders transporter-CYP450 interplay via altering expressions and functions of transporters and CYP450s, finally affecting pharmacokinetics and activity/toxicity of drugs. For example, diabetes upregulated expressions and functions of hepatic CYP3A and OATP1B2, enhanced hepatic uptake and metabolism of simvastatin [12], atorvastatin [9,11], in turn, increasing hepatoxicity of atorvastatin [76]. Data from PBPK also showed that both hepatic OATP1B2 and CYP3A contribute to the clearance of atorvastatin, but the roles of hepatic OATP1B2 were much larger than that of CYP3A [11]. Another example is cyclosporin A. Cyclosporin A is also a substrate of CYP3A, MRP2, P-gp, OATP1B1 and BCRP. Significantly increased systemic clearance of cyclosporin A in diabetic rats [77] should be attributed to the net effect of alterations in these transporters and CYP3A. Methotrexate transport in the liver is mediated by various transporters such as OATP1B1, MRP2, BCRP and OAT2. Moreover, induction of CYP2E1 also enhanced methotrexate-induced hepatocytoxicity [78]. These results indicate that the regulation of hepatic OATPs and CYP2E1 expressions and functions may explain clinic findings that diabetic patients were particularly at increased risk of methotrexate hepatotoxicity [79].

#### **3. Intestine**

A series of transporters (such as P-gp, BCRP, OATPs, OCT1, PepTs, MRP2, MRP3 and MCTs) and enzymes (such as CYP450s and UGTs) have been expressed in enterocytes, implicating intestinal absorption of drugs and first-pass effects (Figure 2).

**Figure 2.** (**A**) Possible location of main transporters in human intestine and (**B**) roles of transporter-CYP3A interplay in disposition of atorvastatin in enterocytes. Symbol: ATR, atorvastatin; BCRP, breast cancer resistance protein; MCTs, monocarboxylate transporters; MRPs, multidrug resistance-associated proteins; OATPs, organic anion transporting polypeptides; OCTs, organic cation transporters; PepT1, peptide transporters; P-gp, P-glycoprotein; OH-ATR, hydroxyl atorvastatin; ATR-Glu, atorvastatin acyl glucuronide.

#### *3.1. P-gp*

P-gp is highly expressed at the apical membrane of the intestinal epithelium, pumping out of its substrates from enterocytes to the intestinal lumen. The expression of intestinal P-gp protein is region-dependent. P-gp protein progressively increases from proximal to distal regions. The highest expression of P-gp occurs at the ileum [80]. Several studies have demonstrated that diabetes impairs expression and function of intestinal P-gp, leading to the enhancement of intestinal absorption and increasing plasma exposure following oral administration of P-gp substrates such as arctigenin [81], protoberberine alkaloids [82], digoxin [83], grepafloxacin [84], paclitaxel [17] and morphine [85]. In line with the increased plasma levels of morphine, diabetic mice showed a stronger analgesic effect following an oral dose compared with control mice [85]. The decreased expression and function of intestinal P-gp under diabetes status may explain that clinic findings that diabetic patients also showed higher serum concentration of digoxin compared with non-diabetic patients following an oral dose of digoxin [86,87].

Downregulation of intestinal P-gp expression by diabetes was considered to be partly attributed to the acceleration of the ubiquitin-proteasome [88] via nitric oxide synthase (NOS) activation [89–91]. However, the role of NO in intestinal P-gp is time-dependent. In Caco-2 cells, it was reported that short-term exposure to sodium nitroprusside impaired P-gp function and expression, whereas long-term exposure stimulated P-gp function and expression [92]. Increased levels of short chain fatty acids (SCFAs) were also found in the intestinal content of diabetic rats [47]. A recent study showed that SCFAs downregulated the expression of intestinal P-gp via inhibiting histone deacetylase and NF-κB pathways [93].

#### *3.2. MRP2*

MRP2 (and MRP1) is co-located with phase II conjugating enzymes (such as UGTs and glutathione *S*-transferase) in the small intestine, pumping out of anions from enterocytes to the intestinal lumen. In contrast to P-gp, diabetes was reported to enhance the expression and function of intestinal MRP2 in rats [23]. Gliclazide is also a substrate of MRP2 [94]. Cmax and AUC0–tn of gliclazide following oral administration to diabetic rats were only 55% and 56% of those in control rats [94], respectively, which seemed to be attributed to the increased expression of intestinal MRP2. As expected, gliclazide administration significantly decreased glucose concentrations in healthy rats, but hypoglycemic effects of gliclazide were not observed in diabetic rats [95].

#### *3.3. BCRP*

Similarly to P-gp, BCRP is expressed at the luminal membrane of enterocytes. The BCRP protein gradually rises from proximal to distal regions. Diabetes-induced alterations in the expression of intestinal BCRP are dependent on the type of diabetes and duration of diabetes. STZ-induced diabetic rats showed significantly lowered expression and function of intestinal BCRP, accompanied by decreased intestinal clearance of glibenclamide and increased apparent intestinal effective permeability (Peff) [48], leading to the increased oral plasma exposure of glibenclamide following oral administration to diabetic rats. However, in STZ/HFD-induced diabetic rats, 10-day diabetic rats showed significantly lowered expression and function of intestinal BCRP, but remarkable inductions of intestinal BCRP expression and function were found in 22-day diabetic rats [11]. The induction of intestinal BCRP was considered to be related to the increased levels of short chain fatty acids (SCFAs) via activating PPARγ [96].

#### *3.4. PepT1*

PepT1, located on the luminal membrane of enterocytes, is responsible for intestinal absorption of peptidomimetic drugs, dipeptides and tripeptides resulting from the dietary breakdown of proteins and bacterial peptidomimetics. Several investigations have demonstrated that diabetes downregulated the expression and function of intestinal PepT1 in experimental animals [47,97–99], decreasing oral plasma exposures of cephalexin and acyclovir following oral dose of cephalexin and valacyclovir [99]. Importantly, altered expression and function of intestinal PepT1 are also dependent on sex. For example, in Sprague-Dawley rats, it was reported that STZ-induced diabetes decreased the expression and function of intestinal PepT1 in male rats, but increased them in female rats [100]. Leptin and insulin

deficiencies may contribute to the downregulation of intestinal PepT1 [101,102]. Dyshomeostasis of bile acid compositions was observed in intestinal content of diabetic rats. Bile acids downregulated the expression of intestinal PepT1 via activating farnesoid X receptor (FXR) [99].

#### *3.5. MCT6*

MCT6, located on the apical side of human intestinal villous epithelial cells, mediates intestinal absorption of nateglinide [103]. Loop diuretics such as furosemide, piretanide, azosemide, and torasemide may also be substrates of MCT6 [104]. Diabetes significantly decreased the expression of intestinal MCT6 and decreased plasma exposure of nateglinide following oral dose to rats. Further study demonstrated that SCFAs, especially butyrate, downregulated the expression of intestinal MCT6 via activating PPARγ [47]. The downregulation of intestinal MCT6 may also partly explain that diabetes significantly decreased plasma exposures of furosemide [105] and azosemide [106] following oral doses to rats.

#### *3.6. CYP450s*

Intestine also highly expresses CYP450s, especially CYP3As, which mediate drug metabolism, contributing to first-pass effects. In STZ-induced diabetes, it was found that diabetes significantly decreased CYP3A activity. In line with this, formations of norverapamil [27] and 6-β hydroxyltestosterone [107] were reduced to 21% and 50% that of control rats, respectively. Significant downregulation of intestinal CYP3A expression and function was also found in STZ/HFD-induced diabetic rats [11] and in Tsumura Suzuki obese diabetic (TSOD) mice [108]. Insulin treatment partly reversed the decreased CYP3A expression in STZ-induced diabetic rats [107], but not in TSOD mice [108]. However, the real mechanism leading to the decreased expression of intestinal CYP3A was unclear. In generally, expressions of CYP3A and P-gp were mainly controlled by pregnane X receptor (PXR). We recently reported upregulation of intestinal FXR [83]. A report showed that FXR agonist GW4064 remarkably induced expression of small heterodimer partner (SHP) [109], in turn, inhibiting the transcriptional activity of PXR [109,110]. Moreover, levels of bile acids, FXR agonist, were increased in intestinal contents of diabetic rats, which inhibited the expression of intestinal CYP3A and P-gp via the activation of FXR-SHP pathway.
