Advances in plant-based inhibitors of P-glycoprotein

Abstract

Multidrug resistance (MDR) has emerged as the main problem in anti-cancer therapy. Although MDR involves complex factors and processes, the main pivot is the expression of multidrug efflux pumps. P-glycoprotein (P-gp) belongs to the family of adenosine triphosphate (ATP)-binding cassette (ABC) transporters. It functions in cellular detoxification, pumping a wide range of xenobiotic compounds out of the cell. An attractive therapeutic strategy for overcoming MDR is to inhibit the transport function of P-gp and thus, increase intracellular concentration of drugs. Recently, various types of P-gp inhibitors have been found and used in experiments. However, none of them has passed clinical trials due to their high side-effects. Hence, the search for alternatives, such as plant-based P-gp inhibitors have gained attention recently. Therefore, we give an overview of the source, function, structure and mechanism of plant-based P-gp inhibitors and give more attention to cancer-related studies. These products could be the future potential drug candidates for further research as P-gp inhibitors.

• Multidrug resistance
• P-glycoprotein
• P-glycoprotein inhibitors
• plant-based medicines

Introduction

Some cancer patients can be cured by chemotherapy, immunotherapy and biological-response modifiers, but others respond transiently or do not respond at all to these approaches1–4. Despite recent advances in cancer drug therapy, most cancer patients eventually end up in treatment failure5. Host and tumor-genetic alterations, epigenetic changes and tumor environment, all seem to contribute to the complex process of cancer drug resistance, i.e. the main cause of drug failure is drug resistance1.

Multidrug resistance (MDR) is described as a phenomenon by which resistance to drugs is accompanied by resistance to other drugs whose structures and mechanisms of action may be completely different6. In tumor7,8, MDR is usually associated with an adenosine triphosphate (ATP)-dependent decrease in cellular drug accumulation, which is attributed to the over-expression of certain ATP-binding cassette (ABC) transporter proteins9.

P-glycoprotein (P-gp), the key member of the MDR family proteins, was first characterized in MDR Chinese hamster ovary (CHO) cells by Ling et al.10. It is reported to be the best characterized efflux pump that mediates MDR, and it belongs to the ABC protein superfamily11. The function of P-gp is to prevent exogenous substances from accumulating in cells by transporting a wide variety of structurally and functionally unrelated compounds from the cell interior into the extracellular space. The energy required for the efflux is supplied by ATP hydrolysis12.

P-glycoprotein inhibitors, the reversal of MDR through direct or indirect interaction with P-gp, are classified into three generations. The classification is based on their specificity, toxicity and affinity13. The first-generation inhibitors have been used in clinics, and exhibited inherent pharmacological activities14. These include verapamil15, nifedipin16, cyclosporin A17, ibuprofen18, reserpine19, amiodarone20 and tamoxifen21 (Table 1). However, the usage of these compounds is limited by their toxicity due to the high serum concentrations achieved with the dose required to inhibit P-gp. A greater deal of research by academicians in the direction of decreasing potential toxicity profile and improving specificity resulted in the second- and third-generation inhibitors. Second-generation inhibitors have decreased pharmacological activity but increased affinity than the first-generation inhibitors22. However, inhibition of two or more ABC transporters leads to complicated drug–drug interactions by this class of compounds. These inhibitors include non-immunosuppresive analogs of cyclosporin A, valspodar23; d-isomer of verapamil, dexverapamil24. Others are MM3625, cinchonine26 and quinine homodimer Q227 (Table 1). In order to avoid the complications associated with second-generation inhibitors, novel third-generation inhibitors were developed. This generation of P-gp inhibitors was shown to be more specific for P-gp; hence, avoiding interference with the pharmacokinetics (bioavailability, excretion, etc) of other drugs13. These P-gp inhibitors include but not limited to tariquidar (XR9576)28, zosuquidar (LY335979)29, elacridar (GF120918)30 and PGP400831 (Table 1). Although proven to be better than second-generation inhibitors, third generation of P-gp inhibitors could not “stand the heat” in clinical studies. For example, the XR9576 was stopped in Phase II trial, because a higher proportion of adverse events were reported in its clinical research32. This suggests that the search for more effective P-gp inhibitors is undoubtedly needed.

Table 1. Three classifications of P-gp inhibitors.

Generations	Examples	Limitations
First-generation inhibitors	Verapamil, nifedipin, cyclosporin A, ibuprofen, reserpine, tamoxifen amiodarone	These compounds have inherent pharmacological activities, non-specificity and low affinity.
Second-generation inhibitors	Valspodar, dexverapamil, MM36, cinchonine and quinine homodimer Q2	Inhibition of two or more ABC transporters leads to complicated drug–drug interactions by this class of compounds.
Third-generation inhibitors	Tariquidar (XR9576), zosuquidar (LY335979), elacridar (GF120918) and PGP4008	These compounds can inhibit P-gp with high specificity and selectivity. However, comprehensive clinical trials are warranted.

Plant-based medicines have long been used in the treatment of tumors33. They are reported to have numerous advantages which include but not limited to low cost, lower toxicity and a wider number of targets. Several plant-based compounds and plant extracts have been reported to inhibit P-gp and modulate MDR34. Based on basic structure, plant-based compounds are classified into flavonoids, alkaloids, coumarins, terpenoids and sterides.

Therefore, in this review, we will discuss these compounds and their sources, functions, structure and mechanism in relation to their P-gp inhibitory activities, giving more attention to cancer-related studies.

P-glycoprotein

Structure, distribution and function

P-glycoprotein consists of 1280 amino acids and has two homologous halves, each containing a transmembrane domain (TMD1 and TMD2), which contain six membrane spanning (TM) segments (TM1–6 on TMD1, and TM7–12 on TMD2), and two nucleotide binding domains, NBD1 and NBD2, which includes an ATP site, located in the cytosol. Also, there are three glycosylation sites at the first extracellular loop in the N terminal35–39 (Figure 1).

Figure 1. Predicted secondary structure of P-gp and different mechanisms of P-gp inhibition. TMD – transmembrane domain; TM – membrane spanning; NBD – nucleotide binding domains; the ATP sites are rectangles; the three black lines represent glycosylation sites.

The expression of P-gp was identified in a wide range of non-neoplastic tissues in humans and other mammalian organs40–42. These organs include liver43, pancreas44, kidney45, intestine46, adrenal gland47 and brain48 (Table 2).

Table 2. P-gp expression in different organs.

Organ	Site of expression	Reference
Liver	Biliary canalicular front of hepatocytes.	43
Pancreas	Apical surface of the epithelial cells of small ductules.	44
Kidney	Proximal tubules and mesangial cells.	45
Colon and jejunum	Apical superficial columnar epithelial cells.	46
Adrenal gland	Surface of cells in both the cortex and medulla.	47
Brain	Apical membrane of choroid plexus epithelial cells.	48

Due to selective distribution at the port of drug entry and exit, P-gp has been speculated to play a major physiological role in absorption, distribution and excretion of xenobiotics49. P-gp mainly functions as exogenous substances entry barrier and as a vacuum cleaner that expels xenobiotics from the brain, liver, kidney, intestine, etc.

Mechanism of P-glycoprotein efflux

The exact mechanism by which P-gp combines ATP hydrolysis with the movement of xenobiotics across the plasma membrane, to and out of cell and the exact site of substrate interaction with the protein, are not well defined50. However, there are various models which provide clues that can explain the mechanism of P-gp. There are three prevalent models, namely, the pore model, flippase model and hydrophobic vacuum cleaner (HVC) model. The HVC model combines the features of pore and flippase models. In HVC model, P-gp can identify the parcel on the inside of the cell membrane substrate molecule and transfer them through protein channels to the serous lateral51. These models assumed that P-gp substrates divide into the lipid phase prior to interacting with the protein, which would help to explain the P-gp broad substrate specificity.

There are also different models to explain the translocation process of P-gp. But, all models can be summarized into four distinct stages as follows: (i) drugs and one molecule of ATP combine with TMD and NBD dimerization, (ii) reorientation of the TMD from high to low affinity, (iii) hydrolysis of ATP into adenosine diphosphate (ADP) and subsequent production of energy for drugs excretion from the membrane and (iv) termination of the cycle when ADP fell from P-gp52. The next catalytic cycle begins with another molecule of ATP hydrolysis, which provides energy for TMD reset.

Mechanism of P-glycoprotein inhibitors

P-glycoprotein substrates are capable of binding to P-gp specifically or non-specifically53. Drugs, such as antibiotics, alkaloids, peptides, steroid hormones and local anesthetics are all substrates of P-gp that are widely used in the clinics54. The physicochemical properties of these substrates are not the same. Most of them are strongly lipophilic, often contain aromatic rings and passively diffuse through cell membrane55. Some P-gp substrates are capable of inhibiting the efflux pump function of P-gp or reversing the MDR of tumor cells56. The therapeutic use of P-gp inhibitors to improve drug bioavailability in tumor chemotherapy by inhibiting P-gp in the intestine, liver and kidney, has gained considerable interests49. In fact, since the recognition that P-gp-mediated drug resistance is clinically important, a concerted effort to screen for P-gp inhibitors has extensively dominated research57. The following sections, throw more light on the mechanisms of P-gp inhibitors.

P-glycoprotein inhibitors can inhibit the function of P-gp through three main mechanisms (Figure 1): (I) Inhibition using either the competitive, non-competitive or allosterism drug binding sites56,58. Examples include the drugs such as, competitive inhibitors; vinblastine59, nitrendipine59 and prazosin60; non-competitive inhibitors; amoxapine22 and loxapin22, and allosteric sites; verapamil59. (II) By interfering with ATP hydrolysis61. For example, quercitin, a naturally occurring flavanoid, has been proposed to block P-gp function by interfering with ATPase activity62. (III) By changing the integrity of cell membrane lipids, so that the inhibitors can interfere with membrane fluidity. These types of P-gp inhibitors are pharmaceutical excipients that can change the integrity of cell membrane lipids63. Examples include poly(ethylene glycol)-300, Tween-80, Cremophor EL, etc63.

Plant-based P-glycoprotein inhibitors

Flavonoids

Flavonoids are a group of natural substances that possess a 2-phenylchromone with substituent groups attached at position 2, 3 or 4, and found in fruits, grains, bark, roots, stems, flowers, tea and wine64,65. Based on different substitution, attached position of ring B and the oxidation status of ring C, flavonoids are divided into many sub-classes: chalcones, flavones, flavonols, flavanones, flavanols, anthocyanins and isoflavones65. Many plant extracts that are rich in flavonoids have been proved capable of modulation activity and expression of P-gp66.

For instance, Zuccagnia punctata (ZpE) Cav. (Fabaceae), is a monotypic species widely distributed in western Argentina. ZpE and two of its components: 3,7-dihydroxyflavone (DHF) (1) and 2′,4′-dihydroxychalcone (DHC) (2) (Figure 2, Table 3) have been reported to be modulators of both the expression and activity of P-gp. The rhodamine123 assay, an indicator to show the effects of different treatments on the efflux of P-gp, performed on the human proximal tubule cell line (HK-2) exposed to DHF and DHC shows a significant dose-dependent increase of intracellular fluorescence. Furthermore, the IC₅₀ values for the inhibitory effects on P-gp activity by DHF and DHC were 3.2 and 6.0 μg/mL, respectively67.

Figure 2. The structure of flavonoids that can inhibit P-gp. A: the basic flavonoids structure.

Table 3. An overview of flavonoid inhibitors from in vitro studies.

Compound	P-gp inhibition	Cell line	Reference
1	IC50 = 3.2 μg/mL	HK-2	67
2	IC50 = 6.0 μg/mL	HK-2	67
3	At 30 μM, IC50 = 1667 ± 240 nM	KB-V1	68
4	At 10-50 μM	LS 180	69
5	At 200 μM	MCF-7	70
6	At 30 μM, IC50 = 1233 ± 202 nM	KB-V1	68
7	At 15 μM, IC50 = 0.238 ± 0.085 μM	HepG2/ADR	71
8	At 10–30 μM	MCF-7/ADR	72
9	At 100 μM, IC50 = 0.80 ± 0.20 μM	MDA435/LCC6MDR1	73
10	At 100 μM, IC50 = 8.74 ± 5.88 μM	MDA435/LCC6MDR1	73
11	At 30 μmol/L	HL-60	74

IC₅₀= the concentration required to achieve 50% inhibition of cell growth; HK-2 = human proximal tubule cell line; KB-V1 = human cervical carcinoma cell line; LS 180 = human colon adenocarcinoma cell line; MCF-7 = human breast cancer cell lines; HepG2/ADR = human hepatocellular carcinoma/adriamycin cell line; MCF-7/ADR = doxorubicin resistant breast cancer cell line; MDA435/LCC6MDR1 = human breast cancer cell line; HL-60 = human leukemia cell line.

Another compound, quercetin 3 (Figure 2, Table 3), isolated from Ginkgo biloba leaves has been shown to strongly inhibit P-gp mediated transport of fluorescent probe Hoechst 33342, a substrate of P-gp, by inhibiting the ATPase activity up to about 33% with 25 μM quercetin75,62. Shapiro and Ling62 hypothesized that the effect of quercetin on drug efflux from MDR CH^RC5 cells (Chinese hamster ovary cells) has no direct effect on P-gp. Futhermore, Limtrakul et al.68 reported that treatment with 30 μM quercetin was able to significantly decrease P-gp level in MDR human cervical carcinoma cell line (KB-V1).

To add, rutin 4 (Figure 2, Table 3) was isolated from Ruta graveolens grass naphtha76. Zhang et al.76 reported that rutin decreased the bioavailability of cyclosporine through activation of P-gp. Thus, P-gp is involved in the transmembrane transport and intracellular accumulation of rutin in human colonic cancer cells (Caco-2). Yu and co-researchers69 have found that rutin significantly decreased the accumulation of rhodamine 123 in LS 180, a human colon adenocarcinoma cell line, indicating a possible activation of P-gp.

Additionally, genistein (GNT) 5 (Figure 2, Table 3) is ingested as a component of vegetables (e.g. soy beans) or as a dietary supplements77. Castro and Altenberg70 reported that GNT inhibited rhodamine123 efflux in human breast cancer cell lines (MCF-7). GNT also decreased photo-affinity labeling of P-gp with [³H]azidopine, a P-gp substrate, suggesting that GNT could block P-gp-mediated drug efflux by direct interaction with P-gp.

Similarly, kaempfero 6 (Figure 2, Table 3) from Kaempferia galanga L. root caused a decrease in P-gp levels in human glioblastoma cell line (T98G). A study by Nakatsuma and colleagues78 revealed that kaempferol reduced P-gp expression and function resulting in the inhibition of P-gp activity. Limtrakul et al.68 also reported that treatment with 30 μM kaempfero was able to significantly decrease the P-gp level KB-V1 cells.

In a similar study, icaritin 7 (Figure 2, Table 3), a hydrolytic product of icariin, was isolated from Herba epimediu. Icaritin significantly increased the intracellular accumulation of adriamycin (ADR) and decreased the expression of MDR1 gene in HepG2 (a human hepatocellular carcinoma (HCC) cell line)/ADR cell line. Although the HepG2/ADR cell line was developed by treating the cells with ADR only, it was observed that MDR was achieved when assessed with other anti-cancer drugs, such as vincristine (VCR), cisplatin and 5-fluorouracil. Futhermore, the icaritin-mediated reversal of HepG2/ADR cell resistance to ADR was observed at a concentration of 1, 15 and 30 μM71.

In another study, baicalein 8 (Figure 2, Table 3) was isolated from the roots of Scutellaria. Li et al.72 found that oral administration tamoxifen (10 mg/kg) to rats in the presence of baicalein (0.5, 3, and 10 mg/kg) significantly increased the oral bioavailability of tamoxifen from 47.5 to 89.1%. This was attributed to inhibition of the CYP3A (cytochrome P450, family 3, subfamily A) – mediated metabolism of tamoxifen and inhibition of the P-gp efflux pump in the small intestine.

Last but not the least, biochanin A 9 and silymarin 10 (Figure 2, Table 3) were isolated from the bark of Aesculus hippocastanum L. and seeds of Milk Thistle, respectively. Zhang and Morris79,73 reported that at 1.5 h, accumulation of digoxin and vinblastine in Caco-2 significantly increased and the AP-to-BL (from apical side to basolateral side in Caco-2 tranwell assay) transport of digoxin also significantly increased with 50 μM biochanin A or silymarin. This indicates that biochanin A and silymarin can inhibit P-gp-mediated efflux in Caco-2 cells, and may potentially increase the absorption/bioavailability of co-administered drugs that are P-gp substrates.

Lastly, wogonin 11 (Figure 2, Table 3), was extracted from the roots of Scutellaria baicalensi Georgi. Wogonin was reported to significantly potentiate etoposide-induced apoptosis in human leukemia cell line (HL-60). This effect was attributed to impairment of P-gp function and subsequent increase in cellular content of etoposide in the cells. The potentiation by wogonin is likely to be a specific action for cancer cells, which was also observed by Lee et al.74. Consequently, it is suggested that wogonin may be used to reduce the excretion of the anti-cancer agents via P-gp and enhance their pharmacological actions in cancer cells.

Structural features of flavonoids are found to contribute positively to P-gp inhibition. For instance, it is reported that a hydroxyl group in position 5, double bond between positions 2 and 3, and a methoxyl group in position 3 are required for their P-gp inhibitory activities. Also, the exchange of a 3-methoxy group by an OH group can decrease their activity80. This structural-activity relationship of flavonoids has also been reported by other researchers. For instance, Boumendjel et al.66 have reported that 3-prenylchalcone binds to NBD2 with 20-fold higher affinity compared to the unprenylated chalcone.

As discussed above, flavonoids have significant inhibitory activities on expression of P-gp in various tumor models. While herb–drug interactions may set potential limits flavonoids co-administered with other drugs, opportunities exist for the combination of flavonoids with suitable anti-cancer drugs to improve therapeutic index and bioavailability of such anti-cancer drugs and reduce the dose to minimize side effects associated with anti-cancer drugs81. Also, bioavailability of flavonoids can be influenced by factors, such as pH of the gastrointestinal tract82, low hydrophobicity83 and CYP enzymes84; hence, future scientific research to overcome these challenges are needed to explore the vast pharmacological activities of plant-derived flavonoids.

Alkaloids

Alkaloids are natural nitrogen-containing secondary metabolites found in plants, fungi and bacteria85. The vast majority of alkaloids are reported to be present in higher plants especially in gymnosperms and angiosperms and a few exist in lower plants. A fundamental characteristic of alkaloids is the possession of nitrogen, which can be derived from amino acids or basic structure of purine or amination of sterides and terpenes85. Many researchers have reported on the modulatory effects of alkaloids on P-gp, and so we discuss these effects as follows.

Glaucine 12 (Figure 3, Table 4), is an isoquinoline alkaloid, isolated from the stems of Corydalis yanhusuo. Glaucine inhibits P-gp mediated efflux and activates ATPase activities of transporters, indicating that it is a substrate of and can competitively inhibit P-gp. Furthermore, glaucine has been shown to decrease IC₅₀ values of the MCF-7/ADR cells to adriamycin and mitoxantrone in a dose-dependent manner (12.5, 25 and 50 μM)86.

Figure 3. The structure of alkaloids that can inhibit P-gp. B: the basic alkaloids structure.

Table 4. An overview of alkaloid inhibitors from in vitro studies.

Compound	P-gp inhibition	Cell line	Reference
12	Adriamycin + 25 μM 12, IC50 = 8.86 ± 0.08 μg/mL	MCF-7/ADR	86
13	Doxorubicin + 10 μM 13, IC50 = 0.35 ± 0.03 μg	Caco-2	87
		Doxorubicin + 10 μM 13, IC50 = 19.70 ± 3.91 μM		CEM ADR5000
14	IC50 = 6.6 μg/ml	KB-V1^+	88
15	At 2–5 μM	K562	89
16	IC50 = 27.68 ± 1.04 μM	MCF-7	90
17	At 3 μM	HCT15	91
18	At 3 μM	HCT15	91
19	At 10*10^−3 M	K562/R10	92
20	IC50 = 0.17 μM	KB-V1	93
21	IC50 = 0.17 μM	KB-V1	93
22	Vinblastine + 5 μM 22, IC50 > 2 M	KB-V1	94

IC₅₀= the concentration required to achieve 50% inhibition of cell growth; MCF-7/ADR = doxorubicin resistant breast cancer cell line; Caco-2 = human colonic cancer cells; CEM ADR5000 = P-gp over-expressing sub-line of CCRF-CEM; KB-V1⁺= MDR human epidermoid carcinoma assessed in the presence of vinblastine (1 μg/mL); K562 = human chronic myelogenous leukemia cell line; MCF-7 = human breast cancer cell line; HCT15 = colorectal cancer cell line; K562/R10 = human leukemia cell line; KB-V1 = human cervical carcinoma cell line.

Lobeline 13 (Figure 3, Table 4), a piperidine alkaloid from Lobelia inflata, has been reported to specifically inhibit P-gp activity. It reverses doxorubicin resistance in Caco-2 and CEM ADR5000 [a P-gp over-expressing sub-line of CCRF-CEM (the human acute lymphocytic leukemia T lymphocytes)] in vitro at non-toxic concentration, at 10 μM87.

In another study, tropane alkaloid esters were isolated from the stems of Erythroxylum rotundifolium. One of them is 7a-hydroxy-6a-(3,4,5-trimethoxybenzoyloxy)-3R-(E)-(3,4,5-trimethoxycinnamoyloxy) tropane 14 (Figure 3, Table 4). This compound has been demonstrated to have a significant biological activity on KB-V1 cells incubated in the presence of vinblastine. Thus, tropane esters of this type can reverse the MDR phenotype, presumably via interaction with P-gp88.

Cepharanthine 15 (Figure 3, Table 4) is a bisbenzylisoquinoline alkaloid extracted from the roots of Stephania cepharantha Hayata95. Ikeda et al.89 showed that cepharanthine is an effective agent for the reversal of resistance in human chronic myelogenous leukemia cell line (K562). It has also enhanced the sensitivity to doxorubicin and VCR, and enhanced apoptosis induced by doxorubicin and VCR in K562 cells.

Further, 1-[(9E)-10–(3,4-methylenedioxyphenyl)-9-decenoyl] pyrrolidine 16 (Figure 3, Table 4) is an aikaloid isolated from Piper boehmeriifolium90. Wang et al.90 evaluated 25 amide alkaloids from Piper boehmeriifolium and 10 synthetic amide alkaloid derivatives for anti-proliferative activity against eight human tumor cell lines, including P-gp-overexpression of MDR KB (KBvin). They found that 3,4,5-trimethoxy substitution in the phenyl ring of piplartine increased selectivity against KBvin. Furthermore, they demonstrated that Piper methysticum root extracts and its major component, the kavalactones, are potent inhibitors of P-gp96.

To add more, tetrandrine (TET) 17 and fangchinoline (FAN) 18 (Figure 3, Table 4) are bisbenzylisoquinoline alkaloids derived from the root of Stephania tetranda91. Choi et al.91 evaluated the MDR-reversal abilities of these two compounds by comparing with verapamil. They found that TET (3.0 μM), FAN (3.0 μM) and verapamil (10.0 μM) reduced the paclitaxel IC₅₀ in HCT15 cells (colorectal cancer cell line, a P-gp over-expression cell line) to about 3100-, 1900- and 410-fold, and enhanced the accumulation rates of rhodamine123 in HCT15 cells (200–250%). Likewise, Sun et al.97 reported that TET and FAN showed a significant synergistic cytotoxic effect in Caco-2 and CEM/ADR5000 cancer cells in combination with doxorubicin. Furthermore, TET and FAN increased intracellular accumulation of rhodamine123 and inhibited its efflux in Caco-2 and CEM/ADR5000 cells. These results suggest that TET and FAN can reverse MDR by increasing the intracellular concentration of anti-cancer drugs, and thus they could serve as lead compounds for developing new drugs to overcome P-gp-mediated drug resistance in clinical cancer therapy.

In other report, indole-3-carbinol (I3C) 19 (Figure 3, Table 4), an indole type alkaloid, is a glucosinolate present in cruciferous vegetables of the genus Brassica92. Western blot analysis of vinblastine-resistant human leukemia (K562/R10) cells showed that I3C down-regulated induced levels of P-gp in resistant cells. The quantification of immunocytochemically stained K562/R10 cells showed 24, 48 and 80% decrease in the levels of P-gp by I3C for 24, 48 and 72 h of incubation, respectively. These results indicate that I3C could be used as a novel modulator of P-gp and may be effective as a dietary adjuvant in the treatment of MDR cancers92.

Other compounds, pervilleines B 20 and C 21 (PB and PC) (Figure 3, Table 4), are two new tropane alkaloid aromatic esters obtained from a chloroform extract of the roots of Erythroxylum pervillei93. They were unable to inhibit KB-V1 cells growth when vinblastine, PB or PC were administered as single agents, but when co-administered with vinblastine, these compounds showed inhibition of up to 77.7%. Equimolar dose of verapamil when tested was less effective98. These suggest that PB and PC are potential inhibitors of P-gp.

Finally on alkaloids, stemona alkaloids were isolated from the roots of Stemona aphylla and Stemona burkillii. There were three isolated alkaloids: stemocurtisine and oxystemokerrine from S. aphylla, and stemofoline 22 (Figure 3, Table 4) from S. burkillii98. Chanmahasathien et al.94 found that stemofoline could increase the P-gp substrates in a dose-dependent manner. Additionally, P-gp ATPase activity was stimulated by stemofoline in a concentration-dependent manner. All these indicate that stemofoline may interact directly with P-gp to inhibit its activity.

Basically, there are two features present in alkaloid compounds that inhibit P-gp: (i) basic nitrogen atom and (ii) small lipophilic molecules with two planar aromatic rings19. Pearce et al.19 have demonstrated that the pendant aromatic ring represented by the benzoyl moiety is an important feature for interaction of reserpine-type inhibitors with P-gp. Also, it is asserted that alkaloid compounds having a pendant benzoyl function under certain conformational constraints most effectively enhance the cytotoxicity of vinblastine and are able to block the binding of ¹²⁵I-NASV (a P-gp substrate) to P-gp in membranes from CEM/VLB₁₀₀ (a human MDR leukemia cell line) cells. In line with these reports, Wang et al.90 demonstrated that (i) 3,4,5-trimethoxy phenyl substitution is critical for selectivity against KBvin, and (ii) in arylalkenylacyl amide alkaloids, replacement of an isobutyl amino group with pyrrolidin-1-yl or piperidin-1-yl significantly improved their activity.

Coumarins

Natural coumarins comprise a wide class of compounds found throughout the plant kingdom99. They are found at high levels in some essential oils, particularly cinnamon bark oil, cinnamon leaf oil and cassia leaf oil. Most coumarins occur in higher plants, with the richest sources being the Rutaceae, Araliaceae and Umbelliferae100. Coumarin is classified as a member of the benzopyrone family of compounds, all of which consist of α-benzopyrone. Based on the position of substitutes and type in α-benzopyrone, coumarins are classified into four main sub-types: the simple coumarins, furanocoumarins, pyranocoumarins and the pyrone-substituted coumarins101. A number of pharmacological activities of coumarins have been reported along with their P-gp inhibitory activity.

For instance, cnidiadin 23 (Figure 4, Table 5), a furanocoumarin, was derived from Tordylium apulum (Apiaceae). It was reported that this compound significantly inhibited the extrusion of the rhodamine123 and the radiolabeled anti-cancer agent [³H]-vinblastine out of MDR1-transfected Madin–Darby canine kidney (MDCK-MDR1) cells, via competitive inhibition of P-gp transport activity102. It is notable that the proportion of killed MDCK-MDR1 cells (normal cell line), reached 93%, when vinblastine was used in combination with cnidiadin at 10 μM. However, when co-treatment with cnidiadin (10 μM)/VCR on the KB/VCR cells, the cells resulted from long-term exposure of the parental sensitive KB human oral epidermis carcinoma cell line to VCR. No cell toxicity was detected after exposure to cnidiadin and cnidiadin sensitized KB/VCR cells103.

Figure 4. The structure of coumarins that can inhibit P-gp. C: the basic coumarins structure.

Table 5. An overview of coumarin inhibitors from in vitro studies.

Compound	P-gp inhibition	Cell line	Reference
23	IC50 = 10 μM	MDCK-MDR1	102
		IC50 = 43.47 ± 0.27 μM	KB/VCR	103
24	At 10 μM	K562/R7	104
25	At 94 μM	L5178	105
26	IC50 = 5.6 μg/mL	MCF7/ADR	106
27	At 0.5 μg/mL	MCF7/ADR	106
28	IC50 = 2–5 nM	K562/D1-9	107

IC₅₀= the concentration required to achieve 50% inhibition of cell growth; MDCK-MDR1 = MDR1-transfected Madin–Darby canine kidney cells; KB/VCR = long-term exposure of the parental sensitive KB human oral epidermoid carcinoma cell line to vincristine (VCR); K562/R7 = it was obtained by prolonged exposure of the human erythroleukemic cell line (K562) cells to doxorubicin; L5178 = the mouse lymphoma cells; MCF-7/ADR = doxorubicin resistant breast cancer cell line; K562/D1–9 = it was obtained by prolonged exposure of the human erythroleukemic cell line (K562) cells to daunorubicin.

In another study, bioguided fractionation from the roots of Citrus sinensis (Rutaceae) led to the isolation and identification of five coumarins, namely, clausarin, suberosin, poncitrin, xanthyletin and thamnosmonin104. Among these compounds, clausarin 24 (Figure 4, Table 5) inhibited P-gp-mediated drug efflux in K562/R7 cells, which in daunorubicin accumulation assay showed significant activity of about 45%, compared with cyclosporin A.

A new sesquiterpene-coumarin ether (5′β, 9′α, 10′α)-7-O-(3α-methoxy-8′ (12′)-drimen-11′-yl)- scopoletin 25 (Figure 4, Table 5) was isolated and characterized from the Me₂CO extract of whole dried plant of Euphorbia portlandica105. Examined for its effects on the reversal of MDR on mouse lymphoma cells (L5178), sesquiterpene-coumarin ether proved to be more active than the verapamil.

Further, two compounds, galbanic acid 26 (from the roots of Ferula szowitsiana) and farnesiferol A 27 (from the roots of Ferula persica) (Figure 4, Table 5) were assessed on the functionality of drug transporter P-gp using a rhodamine123 efflux assay in MCF7/ADR cells106. Compared to verapamil, galbanic acid (5, 10 and 25 μg/mL), significantly inhibited P-gp activity. In inhibition of the P-gp transporter, farnesiferol A (0.5 μg/mL) was more potent than verapamil.

Lastly, GUT-70 28 (Figure 4, Table 5), is a tricyclic coumarin that was extracted from the stem bark of Calophyllum brasiliense. Effects of GUT-70 on P-gp over-expressing MDR K562/D1–9 cells showed that K562/D1–9 cell line was 70 times more resistant to daunorubicin than its parental K562 cell line. The study indicated that GUT-70 may be a useful agent for the treatment of leukemia107.

Conclusively, it is important to note that, some substituents present on the basic coumarin structure, such as the substituents at position C₄ (28), C₅ (24), C₆–C₇ (24, 25 and 28), ether at position C₇ (25, 26, 27 and 28) and a [α,β-di(hydroxyisopropyl)dihydrofuran] group at positions C₇–C₈ (23), exhibit a significant inhibitory effect on P-gp101. For example, Jabeen et al.108 reported that heterocyclic substituents at position C₄ or C₃–C₄ not only enhanced lipophilicity (or log P) but also showed higher inhibitory activity of P-gp. Notably, coumarin itself has low absolute bioavailability in man (< 5%), due to extensive first-pass hepatic conversion of α-benzopyrone to 7-hydroxycoumarin followed by glucuronidation109. It can therefore be proposed that, improving on the bioavailability of coumarins could help enhance their pharmacological activities.

Terpenoids

Terpenoids are a class of isoprenoids from a large and structurally diverse family of natural products. Typical structures contain carbon skeletons represented by (C5) n, and are classified as monoterpenoids (C10), sesquiterpenoids (C15), diterpenoids (C20), sesterterpenoids (C25), triterpenoids (C30), tetraterpenes (C40) and polyterpenes110. Pharmacological activities of terpenoids are extensively researched, particularly, their inhibitory effects on P-gp as discussed below.

(R)-(+)-citronellal 29 (Figure 5, Table 6) is a component of monoterpenoid found in essential oil from Zanthoxyli fructus and present in some edible plants117. Abietic acid 30 (Figure 5, Table 6) is a major component diterpenoid of the rosin fraction of turpentine from pines and other conifers110. Glycyrrhetic acid 31 (Figure 5, Table 6) is found in Glycyrrhizae radix as a major terpenoid and has been used as an anti-allergic and anti-inflammatory drug118. The inhibitory activities of (R)-(+)-citronellal, abietic acid and glycyrrhetic acid on P-gp-mediated efflux of [³H]digoxin at the screened concentration were greater than 50% on LLC-GA5-COL150 cells, the cells were established by transfection of human MDR1 cDNA encoding P-gp into LLC-PK1 derived from porcine kidney. The IC₅₀ values of (R)-(+)-citronellal (167 ± 8 μM), abietic acid (172 ± 9 μM) and glycyrrhetic acid (80.8 ± 5.0 μM) were lower than that of verapamil (237 ± 13 μM) on LLC-GA5-COL150 cells. In vitro model study on Caco-2 cells, showed that these three compounds increased the AP-to-BL transport and decreased the BL-to-AP transport of [³H]digoxin and efflux ratio. Based on these findings, it was concluded that (R)-(+)-citronellal, abietic acid and glycyrrhetic acid could inhibit P-gp-mediated transport111.

Figure 5. The structure of terpenoids that can inhibit P-gp. D: the basic terpenoids structure.

Table 6. An overview of terpenoid inhibitors from in vitro studies.

Compound	P-gp inhibition	Cell line	Reference(s)
29	IC50 = 167 ± 8 μM	LLC-GA5-COL150	111
30	IC50 = 172 ± 9 μM	LLC-GA5-COL150	111
31	IC50 = 80.8 ± 5.0 μM	LLC-GA5-COL150	111
32	At 20 μM, FAR = 12.1	L5178 Y	112,113
				At 60 μM, FAR = 72.9
33	At 20 μM, FAR = 23.1	L5178 Y	112,113
				At 60 μM, FAR = 82.2
34	At 5 μM	K562/R7	114
35	At 40 μg/mL, FAR = 40.3	L5178 Y	115
36	At 40 μg/mL, FAR = 30.7	L5178 Y	115
37	ED50 = 0.028 μg/mL	MES-SA/DX5	116
		ED50 = 0.0011 μg/mL		HCT 15

LLC-GA5-COL150 = the cells were established by transfection of human MDR1 cDNA encoding P-gp into LLC-PK1 derived from porcine kidney; FAR = fluorescence activity ratio, the values were calculated by (MDR treated/MDR control)/(parental treated/parental control); L5178 Y = the human MDR1 gene-transfected mouse lymphoma cells; K562/R7 = it was obtained by prolonged exposure of the human erythroleukemic cell line (K562) cells to doxorubicin; ED₅₀= the concentration required to achieve 50% effect of cell growth; MES-SA/DX5 = human MDR uterine sarcoma cell line; HCT15 = colorectal cancer cell line.

Furthermore, two new macrocyclic jatrophane diterpenes, namely euphomelliferine 32 (Figure 5) and euphomelliferenes A 33 (Figure 5, Table 6), were isolated from methanolic extract of Euphorbia mellifera112.The two Jatrophane diterpenes were evaluated for their effects on the reversion of human MDR1 gene-transfected mouse lymphoma cells (L5178Y) and on human colon adenocarcinoma (COLO 320), using the fluorescence activity ratio (FAR, are used to evaluate P-gp modulating potential) values and rhodamine123 exclusion test. The two compounds were able to successfully modulate P-gp efflux activity by reversing the MDR phenotype of MDR cell lines. These results indicate that the two could be promising lead compounds for the development of effective P-gp modulators113.

In addition, macrocyclic lathyrane polyester Euphorbia factor L₁₀ 34 (Figure 5, Table 6) was obtained from the seeds of the caper spurge (Euphorbia lathyris) by gravity column chromatography and was further purified by crystallization114. The interaction of L₁₀ and its acetyl derivative with P-gp was investigated by Appendino et al.114. The detection of dipolar proton–proton correlations between two olefinic protons; the enone proton (H-12) and H-19 and H-8β, and between the oxymethylene protons and H-4 and H-7α showed that L₁₀ assumes a conformation with both the 20-methyl and the 6-oxymethylene syn-oriented on the α-face of the macrocycle. Again, it was shown that L₁₀ could bind to the cytosolic domain of P-gp with high affinity and also strongly inhibits P-gp-mediated daunomycin transport. Interestingly, a 95% increase in intracellular drug accumulation was measured at a 5 μM concentration of L₁₀. Based on these results, it can be assumed that L₁₀ could be a novel chemotype for P-gp inhibitors.

Additionally, two new tetracyclic diterpene polyesters, euphoportlandols A 35 and B 36 (Figure 5, Table 6), were isolated from an acetone extract of Euphorbia portlandica115. The evaluation of the inhibition of P-gp-mediated drug efflux from L5178 Y mouse T-lymphoma parental cells was performed by determining the intracellular accumulation of rhodamine 123. At a concentration of 40 μg/mL, both compounds were found to have inhibitory activity on P-gp115.

Lastly, obacunone 37 (Figure 5, Table 6), is a limonoids isolated from Phellodendron amurense116. Min et al.116 found that obacunone showed significant P-gp MDR inhibition activity in MES-SA/DX5 (human MDR uterine sarcoma cell line) and HCT15 cells with ED₅₀ value of 0.028 μg/mL and 0.0011 μg/mL, respectively.

It is noteworthy that, two features of terpenoids, the carbonyl groups at the C_2α, C₃, C₈ positions (30, 31 and 35), and a hydrophobic substituent at the C₆ position (37), appear to be the key contributors for their P-gp inhibitory activity119.

It is obvious that P-gp inhibitory effects of terpenoids depend on the chemical structure of the compound. For example, it is reported that triterpenes need a particular pi-electron distribution at the C₂₄ and C₂₅ positions and also lipophilic groups for increasing log P. Finally, carbonyl oxygen and dipole moments of terpenes are as well important in binding to membrane lipids and peptides by complex formation via OH, COOH and NH₂ groups of the P-gp120.

Sterides

Sterides are generally found in higher plants and have long been used against various diseases, including inflammation and cancer121. The basic structure of sterides contains cyclopentano-perhydrophenanthrene with many substituents, which includes a hydroxy group at the C₃ position, methyl at the C₁₀ and C₁₃ positions and different side chains at the C₁₇ position122. Based on the substituents at C₁₇ position, sterides are classified into cardiac glycosides, steroid saponins, steroid hormones and others123. Here, in this section, we highlight on plant-derived sterides that possess P-gp modulatory activities.

Paris saponin VII (PSVII) 38 (Figure 6, Table 7) was isolated from Trillium tschonoskii Maxim124. PSVII dose dependently suppressed cell viability, triggered apoptosis and modulated drug resistance of MCF-7/ADR cells. PSVII treatment in MCF-7/ADR cells led to increased TNFR1, TRAIL R1/DR4, TRAIL R2/DR5, FADD expression, and activation of polymerase caspase-8, and 3. In parallel to the alterations, P-gp expression and activity were also reduced124.

Figure 6. The structure of saponins that can inhibit P-gp. E: the basic saponins structure.

Table 7. An overview of steride inhibitors from in vitro studies.

Compounds	P-gp inhibition	Cell lines	Reference
38	IC50 = 4.10 ± 0.49 μM	MCF-7/ADR	124
		IC50 = 9.54 ± 1.02 μM		MCF-7
39	At 5–40 μM	KBV20C	125
40	IC50 = 188 μM	AML-2/D100	126
		IC50 = 137 μM		AML-2/D × 100
41	At 10 μM	K562/R7	127
42	At 10 μM	K562/R7	127
43	At 10 μM	K562/R7	127

IC₅₀= the concentration required to achieve 50% inhibition of cell growth; MCF-7 = human breast adenocarcinoma cell line; MCF-7/ADR = doxorubicin resistant breast cancer cell line; KBV20C = MDR human fibroblast carcinoma; AML-2/D100 = the daunorubicin-resistant acute myelogenous leukemia; AML-2/D × 100 = the doxorubicin-resistant AML-2 subline; K562/R7 = it was obtained by prolonged exposure of the human erythroleukemic cell line (K562) cells to doxorubicin.

In a flow cytometric analysis, ginsenoside Rg₃ 39 (Figure 6, Table 7), was isolated from Panax ginseng125, promoted accumulation of rhodamine 123 in drug-resistant KBV20C cells in a dose-dependent manner. Additionally, Rg3 inhibited [³H]vinblastine efflux and reversed MDR to doxorubicin, COL, VCR and VP-16 in KBV20C cells. Photo-affinity labeling study with [³H]azidopine revealed that Rg3 competed with [³H]azidopine for binding to P-gp. This research demonstrates that Rg3 can compete with anti-cancer drugs for binding to P-gp thereby blocking drug efflux125.

Moreover, protopanaxatriol ginsenosides (PTG) 40 (Figure 6, Table 7) is another steride isolated from Panax ginseng126. In a recent study, daunorubicin-resistant acute myelogenous leukemia (AML-2/D100) was shown to increase the expression of P-gp, but treatment with PTG reversed resistance in the AML-2/D100 sub-line in a concentration-dependent manner. The effect of PTG (100 μg/mL) on drug accumulation of daunorubicin in the AML-2/D100 subline was 2-fold higher than that observed in the presence of verapamil (5 μg/mL) and 1.5 times less than cyclosporine A (3 μg/mL). Beyond this, PTG at 200 μg/mL or more completely inhibited the azidopine photo-labeling of P-gp. All these indicate that PTG has a chemosensitizing effect on P-gp-mediated MDR cells by increasing the intracellular accumulation of drugs through direct interaction with P-gp at the azidopine site126.

In another study, bio-guided fractionation of the roots of Paris polyphylla (Trilliaceae), led to isolation and identification of three saponins 3-O-Rha(1 → 2)[Ara(1 → 4)]Glc-pennogenine 41, gracillin 42 and polyphyllin D 43 (Figure 6, Table 7)127. These three compounds exhibited strong ability to inhibit P-gp-mediated drug efflux in K562/R7 cells, compared to the cyclosporin A, with respective values of 41, 64 and 45%, at 10 μM. On transfected cells study, the results showed that the 3-O-Rha (1 → 2)[Ara(1 → 4)]Glc-pennogenine exhibited the highest ability to modulate P-gp activity (108 and 77% inhibition at 20 and 10 μM, respectively), compared with cyclosporin A. The value of the gracillin and polyphyllin D were about 48 and 32%, 61 and 46% inhibition at 20 and 10 μM, respectively127.

Conclusively, it is important to add that, substituents of sterides that mainly exist at the C₃ position (38–39, 41–53), at the C₁₆–C₁₇ position (38, 41, 42 and 43) and at the C₁₇ position (39, 40), could be important for their P-gp inhibitory effects128. Also, the number, the length and the position of side chains, and the type of sugar moiety could have different effects on the P-gp inhibition by sterides. Notably, the hemolytic activity of some steroid saponins (e.g. compounds 41–43) is a major drawback for their clinical development. Fortunately, Wang et al.129 have reported on the structure-activity relationship between steroid saponin and development of hemolysis. They reported that steroid saponin: (i) with the same aglycone and the same length of sugar chain, the sugar linkage determines the hemolytic potency; (ii) with substituents on the hydroxyl groups in the sugar residues could significantly affect their hemolytic activity; (iii) with changes on the aglycone could also change the hemolytic activity. Their report raises hopes that it is feasible to develop steroid saponins into potent P-gp inhibitors with decreased hemolytic effects.

Conclusions and future prospective

Even though MDR involves complex genetic factors, modern scientific research could accelerate its drug discovery process because each factor could provide a novel drug target5. There have been studies on MDR for the past four decades since the Ling lab used somatic cell genetics to investigate the properties of colchicine resistance in CHO cells and discovered P-gp10. The role of this efflux transporter in drug pharmacokinetics, such as drug absorption and excretion, is increasingly becoming important and well defined. In earlier parts of this review, we have highlighted that P-gp is highly expressed in various tissues, and so it is clear that P-gp inhibition has greater effects on absorption and tissue distribution of drugs.

Many of the plant-based compounds mentioned in this review could provide insights into wide range possibilities of using different methods to develop effective P-gp inhibitors. It is noteworthy that some of the plant-based medicines are reported to be non-specific P-gp inhibitors, which could affect other protein and human enzymatic targets. Therefore, it is important that developing effective P-gp inhibitors in the future should be guided by factors, such as lower side-effect, mechanism and specificity, particularly, blockade in terms of P-gp drug-binding domain inhibition130. Finally, experimental methods and techniques, such as empirical structure-activity relationships (SAR)120, quantitative structure-activity relationships (QSAR)131, 3-dimensional quantitative structure-activity relationships (3DQSAR)132, pharmacophore studies133 and cysteine-scanning mutagenesis134 should also be considered as important features to guide researchers to discover highly selective and potent P-gp inhibitors.

Declaration of interest

The authors declare that they have no competing interests.

This work was financially supported by the National Natural Science Foundation of China (No. 81330081, No. 81302845), The Training Program of Academic and Technical Leaders in Universities of Anhui province (No. 34), The Anhui Province Natural Science Foundation in University (No. KJ2013A158) and Grants for Scientific Research of BSKY (No. XJ201211) from Anhui Medical University.