Discovery of mitochondria-targeting berberine derivatives as the inhibitors of proliferation, invasion and migration against rat C6 and human U87 glioma cells

Shengnan Fu; Yanqi Xie; Jue Tuo; Yalong Wang; Wenbo Zhu; Sihan Wu; Guangmei Yan; Haiyan Hu

doi:10.1039/C4MD00264D

View PDF VersionPrevious ArticleNext Article

DOI: 10.1039/C4MD00264D (Concise Article) Med. Chem. Commun., 2015, 6, 164-173

Show CompoundsShow Chemical TermsShow Biomedical Terms

Discovery of mitochondria-targeting berberine derivatives as the inhibitors of proliferation, invasion and migration against rat C6 and human U87 glioma cells†

Shengnan Fu‡ ^ac, Yanqi Xie‡ ^a, Jue Tuo ^a, Yalong Wang ^a, Wenbo Zhu ^b, Sihan Wu ^b, Guangmei Yan ^b and Haiyan Hu *^a
^aSchool of Pharmaceutical Sciences, Sun Yat-sen University, Waihuan East Road 132, Guangzhou 510006, China. E-mail: lsshhy@mail.sysu.edu.cn; Fax: +86-020-39343118; Tel: +86-020-39343118
^bDepartment of Pharmacology, Zhongshan School of Medicine, Sun Yat-sen University, Zhongshan II Road 74, Guangzhou 510080, China
^cDepartment of Pharmacy, The People's Hospital of ZhongShan, Sunwen East Road 2, Zhongshan 528403, China

Received 19th June 2014 , Accepted 29th September 2014

First published on 30th September 2014

Abstract

This research aims to synthesize lipophilic berberine derivatives and evaluate their antiglioma effects on C6 and U87 cells. The introduction of substituents with various carbon chain lengths on C-13- or C-9-O-position of the berberine scaffold led to the discovery of several potent inhibitors against glioblastoma cells. Derivatives substituted with the carbon chains of moderate length (twelve carbons) displayed improved lipophilicity and the strongest inhibitory effects. Several compounds presented dose-dependent repression against proliferation (IC₅₀, 1.12–6.12 μM) and blocked migration and invasion by over 60% at lower dose levels. Furthermore, preliminary research about the underlying mechanism for the enhanced antiglioma ability indicated that these analogues preferentially localized into mitochondria, inducing the up-regulation of ROS production. Overall, these compounds represent promising candidates to combat glioblastoma and highlight new insight into the antiglioma therapy through interaction with mitochondria.

Introduction

Glioblastoma multiforme (GBM), a grade IV astrocytoma, is the most common and malignant primary brain tumor.¹ Despite multiple therapies, patients suffering from GBM lives typically less than 14 months after diagnosis.² One subtle characteristic of GBM is the highly invasive behavior, conferring glioma cells with the propensity of rapidly infiltrating neighboring normal tissue and making the complete surgical resection impossible.³ Actually, after surgical removal of a glioma, 96% of the recurrent tumors derived from the residual invasive glioma cells arise within 2 cm of the resection cavity⁴ or immediately adjacent to the resection rim.⁵ Moreover, radiotherapy also appears to have reached its practical limit for at least the partly inherent and/or acquired radio-resistance of high-grade glioma.⁶ Moreover, traditional non-selective chemotherapeutics, such as temozolomide (an alkylating agent) and irinotecan (a topoisomerase 1 inhibitor), have been reported. However, the survival advantage brought by these chemicals is accompanied by certain observed severe side effects.⁷ In addition, some low molecular inhibitors of matrix metalloproteases (MMPs) and anti-vascular pharmaceuticals have been investigated to suppress cancer invasion, but they achieved only limited clinical results.⁸

Conclusively, the invasion and migration ability of glioma cells remains the main reason for current unsatisfactory therapies. Indeed, accumulated data substantiate that GBM invasion is regulated and influenced by extremely diversified and overlapping signal pathways instead of a single one.⁹ Therefore, it is extensively suggested that the future successful therapies in the treatment of GBM will target multiple signal transduction cascades, either by one drug that is aiming at the pivotal intersection of several pathways or by a group of agents aiming at multiple signaling pathways.³

Coincidentally, dozens of lethal signaling pathways appear to be a promising therapeutic target, the mitochondrion organelle.¹⁰ Recent studies have described the crucial role that mitochondria, at least partially, play in tumor invasion. The high in vivo expression feature of Bcl-W (a mitochondrial outer membrane anti-apoptotic protein) has been revealed in invasive glioma cells,¹¹ and the depletion of Bcl-W or Bcl-X_L mediated by siRNA renders invasive glioma cells susceptible to cytotoxic-therapy-induced apoptosis.¹² Moreover, research have shown that the active Akt controlling the balance between cell survival and apoptosis¹³ predominantly localizes at the leading edge of migrating cells,¹⁴ suggesting the relationship between tumor invasion and mitochondria. Hurd et al. have efficiently validated the dual effects of intracellular reactive oxygen species (ROS) mainly produced by mitochondria on triggering signal pathways for cell migration and invasion.¹⁵ In addition, the inhibition of heat-shock protein 90 (HSP90), which preferentially localizes in malignant cell mitochondria but not in their normal cell counterparts,¹⁰ will reduce cell migration and invasion by decreasing MMP-9 expression in GBM.⁹ Therefore, it may be possible to target mitochondria to be the junction of multiple signal cascades to achieve improved therapeutic efficacy for GBM patients.

Berberine (1), an isoquinoline alkaloid, possesses many striking pharmacological effects, which have been elaborated in the available literature.¹⁶ Interestingly, berberine can selectively accumulate in tumor cell mitochondria,¹⁷ which could be attributed to its structural amphiphilicity and delocalized positive charge.¹⁸ Moreover, mitochondria-based cancer cell apoptosis induced by berberine has been confirmed in several non-CNS malignant cell lines, with the underlying mechanisms interpreted as decreasing ATP levels and activating apoptotic cascades via lowering the mitochondrial membrane potential, inhibiting complex I of the respiration chain, induction of mitochondrial permeability transition (MPT) and release of pro-apoptotic proteins into cytosol.¹⁹ Furthermore, the recent investigations have started to uncover the potential of berberine in suppressing migration and invasion in non-CNS cancer cell lines. For example, berberine attenuates human gastric tumor cell invasion and inhibits the metastatic potential of breast cancer cells through the down-regulation of MMPs and Akt pathway modulation, respectively.²⁰ However, current literature about whether berberine derivatives could selectively interact with mitochondria and repress the proliferation, migration and invasion of malignant glioma cells is rare. In addition, a significant obstacle that limits the clinical application of berberine is its innate poor solubility and lipophilicity.¹⁶ A structure-activity relationship study on berberrubine derivatives indicated that the presence of the lipophilic alkyl substituents of moderate length presents an optimal antibacterial activity,²¹ prompting us that the lipophilic alteration of berberine may favor its performance in cytotoxicity and anti-migration activity.

Previously, we had evaluated the antiglioma ability of four berberine derivatives varying the alkyl chain length in the C-9-O-position of berberine (3b–e), and found that they showed inhibitory effects on C6 glioma cells in proliferation, migration and invasion.²² In this study, five 13-O-substitued berberine analogs with varying alkyl chain length (5f–j) were synthesized and characterized. However, to comprehensively compare the antiglioma effects of ten berberine derivatives and achieve more reliable results, in other words to avoid time interference on experimental results, the MTT, migration and invasion assays previously used to evaluate the four 9-O-substitued berberine derivatives (3b–e) on C6 cells were performed again. Simultaneously, the same assays of another 9-O-substitued berberine derivative (3a) and the five newly synthesized 13-O-substitued berberine derivatives (5f–j) on rat C6 glioma cells were carried out. In addition, human U87 glioma cells were selected to assess the total ten berberine derivatives in this study. The underlying mechanisms for the improved antiglioma activity were preliminarily explored by studying the subcellular localization of these compounds and measuring ROS production of the C6 and U87 glioma cells with or without berberine derivatives. The BBB penetration ability was also predicted because these compounds have to function inside the brain.

Results and discussion

Chemistry

Synthesis and identification of 3a–e. The synthetic pathway for compounds 3a–e and 5f–j is illustrated in Scheme 1. To obtain 3a–e, the first step was the pyrolysis of 1 to an intermediate berberrubine (2). The following alkylation of the resulting 2 by corresponding alkyl bromides or the benzyl bromide afforded five 9-O-substitued berberine analogs. We found that the key determinant in the synthesis of 9-O-substitued berberine derivatives was to always maintain the reaction system completely free of oxygen. It should be noted that in the alkylation step we simplified the reported condition of reflux at 120 °C ²³ to stirring at 80 °C without compromising the yield. Moreover, the mass spectra of 3a–e show the molecular ions at m/z 462, 490, 546, 574, and 412 (M–Br, ESI-MS), which are 140, 168, 224, 252 and 90 mass units higher than that of the (M + H) of 2, respectively.


	Scheme 1 Synthesis of 3a–e and 5f–j. Reagents and conditions: (a) decalin, N₂, 190 °C. (b) R₁Br, DMF, N₂, stirring, 80 °C. (c) NaBH₄, pyridine, stirring, rt. (d) R₂Br, anhydrous acetonitrile, NaI, N₂, stirring, 80 °C.

Synthesis and identification of 5f–j. To synthesize 5f–j, precursor 1 was first converted into intermediate dihydroberberine (4) by sodium borohydride-promoted reduction. The subsequent sodium iodide-promoted enamine alkylation of 4 by the alkyl bromides of various chain lengths or the benzyl bromide produced five types of 13-substitued berberine analogues. This enamine alkylation step was accompanied by restoring the aromaticity of one nitrogen-containing heterocycle, which is indispensable in maintaining the planar characteristic of the berberine scaffold as well as the pharmacological effects. The ¹H NMR spectra indicate that five compounds 5f–j share the same parent nucleus as 1. The singlet (∼8.95 ppm) displayed by the C-13 position proton of 1 disappears in the ¹H NMR spectra of 5f–j, suggesting that the substitution reaction occurs at C-13 position. Moreover, because of the electron-withdrawing effect of N⁺ at C-7 position, the deshielding and higher field ¹H chemical shifts of the protons in C-13 substituents are also observed in the ¹H NMR spectra of 5f–j, similar to those of 3a–e.

In addition to the above approach for synthesizing 5f–j, we attempted another method involving the reduction of 1 to an intermediate 8-acetonyldihydroberberine, and then alkylation of this intermediate with bromides.²⁴ However, the yield of 8-acetonyldihydroberberine was lower than that of the intermediate 4 because of the former's easy rearrangement to 1 during the heating process. In addition, during the preparation of 4, we compared two reaction solvents, pyridine and the methanol solution containing potassium carbonate, in terms of reaction temperatures, time and yields. Results manifested that the yield of 4 from the reaction with pyridine as the solvent at room temperature for 20 min exceeded that of the reaction with the latter as the solvent at 0 °C for 8 h.

The first obstacle in the synthesis of 5f–i was the considerably low yields that could be ascribed to the following reasons: approximate 30–40% of 4 was converted back into 1 ²⁵ in the enamine alkylation process, which per se possesses the drawbacks of low yields and numerous by-products. The second reason is that alkyl bromides with long carbon chains present poor reaction activity. To improve the yield, we tried to optimize the reaction solvent, pressure, temperature, catalyst, feeding sequence and feeding ratio for enamine alkylation. Firstly, using ESI-MS monitoring the reaction process, we finally selected acetonitrile to dissolve 4 instead of dichloromethane or DMF, both of which exhibited excellent solubility to 4, though the molecular ion peaks of target compounds failed to appear. Secondly, an elevated yield was achieved under high pressure using an airtight thick-walled flask as the reaction apparatus rather than the usual reflux method.²³ Thirdly, we selected 80 °C to be the reaction temperature because no significant improvement of the yield was attained at 100 °C and 130 °C. Fourthly, sodium iodide exhibited higher catalytic activity than potassium iodide and the possible explanation could be the higher solubility of sodium iodide in acetonitrile. Finally, different feeding sequences and ratios were tried and the present sequence and ratio, which are described in the experimental section, were the relatively advantageous combination in the favor of the final yield.

Another obstruction was the purification of 5f–i. Highly pure intermediate 4 was difficult to obtain from either recrystallization or chromatography because 4, as mentioned earlier, easily oxidized back to 1 under aerobic conditions. Consequently, adding impure 4 to the following enamine alkylation produced six by-products per reaction, whose R_f values were very close to those of target products in silica gel-based TLC (Developing solvent: dichloromethane/methanol, 20 : 1, v/v). Neither recrystallization nor silica gel-based column or alumina column chromatography could separate the total resulting mixture.²⁶ After extensive attempts, we developed a three-step approach to remove impurities. Firstly, most of the byproducts were separated by neutral alumina column chromatography (petroleum ether/n-butanol). Secondly, column chromatography using silica gel (n-butanol/water/acetic acid) was harnessed to remove a byproduct whose polarity was closer to that of target product per reaction. Finally, an unidentified impurity peak in ¹H NMR spectra was successfully eliminated after the purification by Sephadex™ LH-20 column chromatography (methanol).

Log P values. In Table 1, the negative log P values of 1, 3e and 5j demonstrate that the introduction of benzyl group to the berberine scaffold at the C-9-O position and C-13 position will moderately augment the lipophilicity of berberine, while alkyl substituents at both C-9-O and C-13 positions of berberine remarkably increase the lipophilicity of berberine. Furthermore, berberine derivatives with moderate carbon chain length (3b and 5g) show the most significant enhancement in lipophilicity, implying more preferable trans-membrane permeability than berberine.

Table 1 Log P values, purity data^a, and effects on the survival rate (IC₅₀, μM) of rat C6 and human U87 glioma cells by 1, 3a–e and 5f–j^b

Compounds	Purity (%)	Log P	IC₅₀ (μM)
Compounds	Purity (%)	Log P	C6	U87
a Log P means the logarithm of the n-octanol/water partition coefficients; purity of these compounds was determined by high-performance liquid chromatography. b IC₅₀: inhibitory concentration causing a 50% reduction in cell proliferation in μM. Data were expressed as mean ± SD.
1	98.89	−1.68 ± 0.24	>20	>20
3a	99.12	2.26 ± 0.34	2.95 ± 0.22	5.73 ± 0.51
3b	99.08	3.43 ± 0.21	1.24 ± 0.13	4.55 ± 0.28
3c	96.23	3.41 ± 0.40	6.12 ± 0.64	15.2 ± 2.02
3d	99.16	3.01 ± 0.40	4.45 ± 0.57	22.49 ±3.38
3e	96.80	−0.86 ± 0.16	11.48 ± 2.05	>20
5f	98.17	2.37 ± 0.27	1.12 ± 0.35	3.65 ± 0.61
5g	96.75	3.66 ± 0.30	1.65 ± 0.38	3.09 ± 0.54
5h	95.62	3.24 ± 0.22	1.67 ± 0.46	3.25 ± 0.27
5i	95.49	3.22 ± 0.31	2.37 ± 0.29	5.75 ± 0.68
5j	99.46	−0.73 ± 0.14	14.4 ± 1.48	>20

Proliferation inhibition in MTT assays. The anti-proliferative effects of compounds 1, 3a–e and 5f–j were assayed in rat C6 and human U87 glioblastoma cells using the MTT ((3, 4, 5-dimethylthiazol-yl)-2, 5-diphenyl tetrazolium) assay that measures the number of metabolically active cells. Fig. 1 illustrates the relative survival rates of both cell lines after incubation with various compounds at different dose levels for 24 h. Firstly, the exposure of C6 cells to 1 with doses up to 20 μM exhibits a minimal anti-proliferative effect, while U87 cells are not sensitive to 1 at this dosage level (Fig. 1A and B). The other derivatives 3a–e and 5f–j are active against both the tumor cell lines in a dose-dependent manner. Specifically, in Fig. 1A and B, 3a and 3b display the strongest anti-proliferation effects on both the cell lines, while 3d and 3e have a moderate effect. In particular, 3c achieves almost the same complete inhibition result against C6 cells as 3a and 3b do when the concentration of 3c reaches 20 μM, whereas the survival rate of U87 cells exposed to 3c is 38.95%. Moreover, for 13-substituted berberine analogs, 5f–i also show robust growth suppression on C6 and U87 tumor cells. For instance, the survival rates are decreased to lower than 20% by 5f–i at 10 μM (Fig. 1C and D). However, 5j exerts a modest repression effect on C6 and U87 cells at 20 μM because the survival rates are as high as 37.24% and 67.20%, respectively. In general, the MTT results of the synthesized derivatives on C6 cells are in similar with those of U87 cells and the C6 cells are more sensitive to all the tested compounds than U87 cells are.


	Fig. 1 Survival rate (%) of C6 (A and C) and U87 (B and D) cells after incubation with different compounds at various concentrations for 24 h (n = 10). Data are presented as mean ± standard deviation. P < 0.05; * versus control; ** versus berberine.

Based on the MTT tests, IC₅₀ values of the assayed compounds were calculated, which are shown in Table 1. As can be seen, 3a–c and 5f–i are superior to the clinically commonly used chemotherapeutics for glioblastoma in terms of cytotoxicity. Indeed, the IC₅₀ values of temozolomide and irinotecan on U87 glioma cells are 19.38 μM and higher than 250 μM, respectively.²⁷ Recently, 3b was synthesized and also proved to be a more robust inhibitor on human cancer HepG2 and HT29 cell lines.²⁸ Moreover, data trends in Fig. 1 and Table 1 demonstrate that the variation of cytotoxic activity of the synthetic analogues correlates with the alternation of alkyl chain length. Namely, the presence of alkyl chains with moderate length (12 carbons) at C-9-O-position or C-13-position on the berberine scaffold guarantees the most appreciable antiglioma activity. However, the introduction of the benzyl group at aforementioned positions inconspicuously contributes to activity elevation. These highly similar varying patterns between the lipophilicity and cytotoxicity of compounds 1, 3a–e and 5f–j led us to postulate that the elevated lipophilicity might promote the transmembrane permeability of these compounds, thus enhancing the intracellular concentrations and cytotoxicity.

Efficacy in transwell migration and invasion assays. As reported, berberine possesses potential repression on migration and invasion in non-CNS cancer cell lines. Therefore, to not only identify, but also to compare the possible suppression of berberine and its lipophilic derivatives on the migration and invasion of glioma cells, we tested 1, 3a–e and 5f–j using transwell migration and invasion assays on Millipore cell culture inserts with 8 μm pore size polycarbonate membrane in C6 and U87 cell lines. Migration and invasion experiments were performed in the absence or presence of matrigel on the polycarbonate membrane, respectively. To rule out cytotoxicity as the driver of decreasing tumor cell migration and invasion ability, first, we selected the minimum highest dose levels of all assayed compounds based on MTT assays so that all the survival rates of all groups were higher than 85%. Second, the number of migrated and invaded tumor cells to the lower chambers was calibrated by dividing by the total cell numbers that were characterized through synchronous MTT assays at the same dose levels.

As shown in Fig. 2–5, all compounds exhibit robust or moderate inhibitory effects against invasion and migration in both the cancer cell lines, except 13-benzyl-subtituted berberine analog 5j. A modest anti-migration and anti-invasion activity is observed in the presence of berberine, which agrees with the previous evidence that berberine could attenuate migration and invasion in other tumor cell lines.²⁰ Furthermore, 3d, 3e, 5h and 5i display greater than 30% suppression and three other potent derivatives (3b, 3c and 5g) block migration or invasion by over 60%. Overall, these experimental results substantiate that the tested compounds are effective migration and invasion inhibitors.


	Fig. 2 Transwell migration (A) and invasion (B) assays in C6 cell lines (n = 3). Dose levels for 1 and 3a–e are 5, 0.5, 0.5, 0.5, 0.5 and 0.5 μM, respectively. Data are presented as mean ± standard deviation. P < 0.05; * versus control; ** versus berberine.


	Fig. 3 Transwell migration (A) and invasion (B) assays in U87 cell lines (n = 3). Dose levels for 1 and 3a–e are 5, 2.5, 2.0, 2.5, 2.5 and 2.5 μM, respectively. Data are presented as mean ± standard deviation. P < 0.05; * versus control; ** versus berberine.


	Fig. 4 Transwell migration (A) and invasion (B) assays in C6 cell lines (n = 3). Dose levels for 1 and 5f–j are 5, 0.5, 0.5, 0.5, 0.5 and 1 μM, respectively. Data are presented as mean ± standard deviation. P < 0.05; * versus control; ** versus berberine.


	Fig. 5 Transwell migration (A) and invasion (B) assays in U87 cell lines (n = 3). Dose levels for 1 and 5f–j are 5, 0.5, 0.5, 0.5, 2.5 and 2.5 μM, respectively. Data are presented as mean ± standard deviation. P < 0.05; * versus control; ** versus berberine.

After comparing Fig. 2A and 3A, it could be concluded that only those analogs with moderate alkyl chain length (3b and 3c) could significantly lower tumor migration tendency. Lengthening the carbon chain (3d) or shortening the substituent (3a and 3e) will compromise the anti-migration potency. Likewise, the similar phenomenon is noted for 5f–j in transwell migration assays (Fig. 4A and 5A). Moreover, there is no cancer type-specific mode of action revealed in migration inhibition evaluation. On the contrary, C6 and U87 cells are differently sensitive to the derivatives regarding invasiveness. For example, the most powerful anti-invasion agents for C6 cells are 3c and 5f, whereas for U87 cells 3a, 3b and 5g are the most remarkable ones (Fig. 2B, 3B, 4B and 5B). Therefore, both the substituents with suitable carbon chain length and tumor cell types influence the activity of synthetic analogues in terms of subduing invasion ability. Intriguingly, agents 1, 3d, 3e and 5j could attenuate cancer cell migration and invasion without exerting acute cytotoxicity. Using these effective but low cytotoxic inhibitors to block tumor cell metastasis will be more advantageous in the long-term treatment of invasive tumor because most of the chemotherapeutics targeting cancer cell proliferation provoke acquired chemoresistance and toxicity to healthy cells after long-time use.²⁹

Subcellular localization in mitochondria. To preliminarily examine the potential target of the synthetic derivatives and if they preferentially localize into mitochondria as berberine does, we selected four compounds (3b, 3c, 5g and 5h) with high anti-growth, anti-migration and anti-invasion ability to incubate with C6 and U87 cells at 1 μM dose level, followed by staining mitochondria with Mitotracker Green FM (500 nM) and photographing. Berberine was also included as a control group. Fig. 6 depicts the laser confocal micrographs of double-stained C6 cells (the left panel) and U87 cells (the right panel).


	Fig. 6 Laser confocal micrographs of double-stained C6 cells (Left panel) and U87 cells (Right panel). Notes: 1. Green channel: 500 nM Mitotracker Green FM stained mitochondria; 2. Yellow channel: 1 μM of respective compounds; 3. Composite images of the former two channels, indicating the co-localization of synthesized derivatives into mitochondria of C6 or U87 cells.

A perfect superposition between Mitotracker Green FM dye images and those of berberine proves the specific subcellular localization of berberine in cell mitochondria, which is in agreement with the previous results.¹⁷ The overlapping photographs for 3b, 3c, 5g and 5h unequivocally demonstrate that these berberine derivatives have a preference for mitochondrial organelles. This type of a mitochondrion-targeting feature could be ascribed to two determinants. The primary reason is that the amphiphilic berberine analogs retain the delocalized cationic charge centers that are crucial for selective mitochondrial accumulation.¹⁸ Secondly, it has been verified that the driving force for this selective accumulation is the mitochondrial transmembrane potential, which proves to be elevated in cancer cells, further favoring the mitochondrial localization propensity.³⁰ In addition, yellow fluorescence intensities of 3b, 3c, 5g and 5h are stronger than that of berberine in both the tumor cell lines. The possible underlying mechanism for the enhanced fluorescence intensities may be that an increase in the lipophilicity from the alkyl substituents promotes the penetration of the synthetic analogs through the cytomembrane, thus improving mitochondrial accumulation.^26b

High generation of reactive oxygen species. It is well established that reactive oxygen species are applied in both cytotoxicity and cancer cell metastasis. Because mitochondria are the main sources of intracellular ROS and the synthesized derivatives appear to preferentially locate in mitochondria, the ROS production of C6 and U87 tumor cells after incubation with or without compounds 1, 3a–e and 5f–j was measured by DHE assay. Cell-permeable DHE is oxidized by O₂⁻ to ethidium bromide, intercalating the cells' DNA and staining the nuclei to be fluorescent red that reflects the ROS levels. In agreement with the varying patterns of lipophilicity (Table 1) and cytotoxicity (Fig. 1), the ROS production of both the cell lines after incubation with 1, 3a–e and 5f–j presented similar changing trends (Fig. 7). For instance, those derivatives (3a, 3b, 5f and 5g) with the alkyl chains of moderate length significantly caused higher ROS generation than that of compounds 1, 3d, 3e, 5i and 5j, whose substituents were either longer or shorter alkyl chains or benzyl group at the C-9-O- or C-13-position on the berberine scaffold. Combined with the mitochondriotropic aspect, these data indicated that the cytotoxicity displayed by the synthesized analogues was derived from the induction of high ROS generation from the mitochondrial organelle.


	Fig. 7 Reactive oxygen species (ROS) generation induced by 2 μM of compounds 1, 3a–e and 5f–j (n = 4). Data are presented as mean ± standard deviation. A: C6 cells; B: U87 cells. ROS levels were determined by the detection of dihydroethidium (DHE) red fluorescence in the nuclei resulting from the oxidation of DHE by intracellular ROS. P < 0.05; versus control; *versus berberine. Con = Control group.

Because mitochondria are the junction of multiple signal pathways responsible for cancer cell proliferation, migration and invasion, the mitochondrion-specific ability and high ROS production tend to be a tentative underlying mechanism for certain derivatives such as 3b, 3c, 5g and 5h, which spontaneously possess anti-proliferation, anti-migration and anti-invasion activity at different dose levels. To further understand the underlying mechanism by which ROS production, which is induced by berberine derivatives, regulates C6 and U87 glioma cells, studies on the mitochondrial membrane potential (ΔΨ_m), release of cytochrome C and apoptosis-inducing factor (AIF), and permeabilization of lysosomes are underway.

Improved BBB penetration of berberine derivatives. To target gliomas, BBB penetration is necessary for these berberine derivatives. As shown in Table 2, berberine only displayed limited BBB penetration potential, while compounds 3b and 5g could readily cross the BBB. Therefore, compounds 3b and 5g fulfil the necessary drug-like and brain penetration criteria. These results demonstrate that introducing the dodecyl group to the C-9-O-position or C-13-position of the berberine scaffold will notably enhance the BBB penetration of synthetic derivatives, favouring their anti-glioma effects.

Table 2 Physical properties and prediction of BBB penetration of 1, 3a, 3b, 5f, and 5g

Calculation^a	MW	c log P	HBA	HBD	PSA	log BB^b
a MW: molecular weight; c log P: calculated logarithm of the octanol–water partition coefficient; HBA: hydrogen-bond acceptor atoms; HBD: hydrogen-bond donor atoms; PSA: polar surface area. b compounds with log BB > 0.3 are able to cross the BBB readily, with log BB <−1.0 are only poorly distributed in the brain.
1	371.87	0.20	4	0	40.82	−0.44
3a	542.51	4.665	4	0	40.82	0.23
3b	570.56	5.675	4	0	40.82	0.39
5f	603.54	5.019	4	0	40.82	0.28
5g	631.59	6.029	4	0	40.82	0.44
Lipinski's rules	≤500	≤5.0	≤10	≤5	≤90	>0.3 (readily cross the BBB); <−1.0 (only poorly distributed to the brain)

Conclusions

Glioblastoma still poses an urgent challenge to both patients and neuro-oncologists. Unfortunately, the management of glioblastoma is mainly palliative all over the world. In this study, an effort was made to develop therapeutics against malignant gliomas through enhancing the lipophilicity of berberine and retaining the mitochondriotropic activity by potentially influencing the intersection of multiple signaling cascades in cancer pathology. Inspired by our previously evaluated 9-O-substitued berberine derivatives on C6 glioma cells,²² 9-O-decyl-berberine (3a) and five cationic C-13-position berberine derivatives (5f–j) varying in the length of substituted alkyl chains were synthesized. The data from biological experiments on C6 and U87 cells suggested that the introduction of the lipophilic substituents with moderate alkyl chain length to the C-9-O-position and C-13-position of the berberine scaffold led to the identification of several potent proliferation, migration and invasion inhibitors in two glioblastoma cell lines. Indeed, the IC₅₀ values for the anti-proliferation of several synthesized compounds reached to low micromolar range (3b, 1.24 μM; 5f, 1.12 μM; C6 cells). Over 60% of invasion and migration was blocked by the derivatives with moderate alkyl chain length (3b and 5g). Furthermore, the simultaneous inhibitory effects against glioma cell survival, migration and invasion should be an integrated result of the improved lipophilicity, the preferential localization into neoplastic mitochondria and the elevated intracellular ROS production. Overall, these compounds provide novel chemotherapeutic candidates and also offer new insight into the antiglioma therapy through the mitochondrial pathway.

Conflict of interest

There is no conflict of interest in this work.

Abbreviations

GBM	Glioblastoma multiforme
MMP	Matrix metalloprotease
Bcl-2	B-cell CLL/lymphoma 2
TLC	Thin-layer chromatography
MTT	(3,4,5-dimethylthiazol-yl)-2,5-diphenyl tetrazolium
IC₅₀	Half maximal inhibitory concentration

Acknowledgements

This study was financially supported by Guangdong Natural Science Foundation (S2012010009237).

Notes and references

D. N. Louis, H. Ohgaki, O. D. Wiestler, W. K. Cavenee, P. C. Burger, A. Jouvet, B. W. Scheithauer and P. Kleihues, Acta Neuropathol., 2007, 114, 97 CrossRef PubMed.
E. G. Van Meir, C. G. Hadjipanayis, A. D. Norden, H. K. Shu, P. Y. Wen and J. J. Olson, Ca-Cancer J. Clin., 2010, 60, 166 CrossRef PubMed.
C. Di, A. K. Mattox, S. Harward and C. Adamson, Anti-Cancer Agents Med. Chem., 2010, 10, 543 CrossRef CAS.
T. Demuth and M. E. Berens, J. Neuro-Oncol., 2004, 70, 217 CrossRef PubMed.
M. E. Berens and A. Giese, Neoplasia, 1999, 1, 208 CrossRef CAS.
A. Deorukhkar and S. Krishnan, Biochem. Pharmacol., 2010, 80, 1904 CrossRef CAS PubMed.
(a) Y. Su, S. Sohn, S. E. Krown, P. O. Livingston, J. D. Wolchok, C. Quinn, L. Williams, T. Foster, K. A. Sepkowitz and P. B. Chapman, J. Clin. Oncol., 2004, 22, 610 CrossRef CAS PubMed; (b) A. B. Schwarzberg, E. H. Stover, T. Sengupta, A. Michelini, M. Vincitore, L. R. Baden and M. H. Kulke, Cancer Invest., 2007, 25, 249 CrossRef CAS PubMed; (c) H. S. Friedman, T. Kerby and H. Calvert, Clin. Cancer Res., 2000, 6, 2585 CAS; (d) S. M. Chang, K. R. Lamborn, M. Malec, D. Larson, W. Wara, P. Sneed, J. Rabbitt, M. Page, M. K. Nicholas and M. D. Prados, Int. J. Radiat. Oncol., Biol., Phys., 2004, 60, 353 CrossRef CAS PubMed; (e) J. J. Vredenburgh, A. Desjardins, J. E. Herndon, J. M. Dowell, D. A. Reardon, J. A. Quinn, J. N. Rich, S. Sathornsumetee, S. Gururangan and M. Wagner, Clin. Cancer Res., 2007, 13, 1253 CrossRef CAS PubMed; (f) T. N. Kreisl, L. Kim, K. Moore, P. Duic, C. Royce, I. Stroud, N. Garren, M. Mackey, J. A. Butman and K. Camphausen, J. Clin. Oncol., 2009, 27, 740 CrossRef CAS PubMed.
S. Zheng, Q. Zhong, Q. Jiang, M. Mottamal, Q. Zhang, N. Zhu, M. E. Burow, R. A. Worthylake and G. Wang, ACS Med. Chem. Lett., 2013, 4, 191 CrossRef CAS PubMed.
J. Drappatz, A. D. Norden and P. Y. Wen, Expert Rev. Neurother., 2009, 9, 519 CrossRef PubMed.
S. Fulda, L. Galluzzi and G. Kroemer, Nat. Rev. Drug Discovery, 2010, 9, 447 CrossRef CAS PubMed.
D. B. Hoelzinger, L. Marianiz, J. Weis, T. Woykeb, T. J. Berens, W. S. McDonough, A. Sloan, S. W. Coons and M. E. Berens, Neoplasia, 2005, 7, 7 CrossRef CAS PubMed.
N. L. Tran, W. S. McDonough, B. A. Savitch, T. F. Sawyer, J. A. Winkles and M. E. Berens, J. Biol. Chem., 2005, 280, 3483 CrossRef CAS PubMed.
X.-M. Yin, Z. N. Oltvai and S. J. Korsmeyer, Nature, 1994, 369, 321 CrossRef CAS PubMed.
A. M. Joy, C. E. Beaudry, N. L. Tran, F. A. Ponce, D. R. Holz, T. Demuth and M. E. Berens, J. Cell Sci., 2003, 116, 4409 CrossRef CAS PubMed.
T. R. Hurd, M. DeGennaro and R. Lehmann, Trends Cell Biol., 2012, 22, 107 CrossRef CAS PubMed.
P. R. Vuddanda, S. Chakraborty and S. Singh, Expert Opin. Invest. Drugs, 2010, 19, 1297 CrossRef CAS PubMed.
G. C. Pereira, A. F. Branco, J. A. Matos, S. L. Pereira, D. Parke, E. L. Perkins, T. L. Serafim, V. A. Sardão, M. S. Santos and A. J. Moreno, J. Pharmacol. Exp. Ther., 2007, 323, 636 CrossRef CAS PubMed.
V. Weissig and V. P. Torchilin, Adv. Drug Delivery Rev., 2001, 49, 127 CrossRef CAS.
Y. Sun, K. Xun, Y. Wang and X. Chen, Anti-Cancer Drugs, 2009, 20, 757 CrossRef CAS PubMed.
(a) J. P. Lin, J. S. Yang, C. C. Wu, S. S. Lin, W. T. Hsieh, M. L. Lin, F. S. Yu, C. S. Yu, G. W. Chen, Y. H. Chang and J. G. Chung, In Vivo, 2008, 22, 223 CAS; (b) H.-P. Kuo, T.-C. Chuang, S.-C. Tsai, H.-H. Tseng, S.-C. Hsu, Y.-C. Chen, C.-L. Kuo, Y.-H. Kuo, J.-Y. Liu and M.-C. Kao, J. Agric. Food Chem., 2012, 60, 9649 CrossRef CAS PubMed.
S. H. Kim, S. J. Lee, J. H. Lee, W. S. Sun and J. H. Kim, Planta Med., 2002, 68, 277 CrossRef CAS PubMed.
Z. G.-g. FU Sheng-nan, T. U. O. JUe, Y. A. N. G. Xiao-xiao, Z. H. U. Wen-bo, Y. A. N. Guang-mei and H. U. Hai-yan, J. Sun Yat-sen Univ., Med. Sci., 2013, 34, 531 Search PubMed.
Y. H. Li, P. Yang, W. J. Kong, Y. X. Wang, C. Q. Hu, Z. Y. Zuo, Y. M. Wang, H. Gao, L. M. Gao, Y. C. Feng, N. N. Du, Y. Liu, D. Q. Song and J. D. Jiang, J. Med. Chem., 2009, 52, 492 CrossRef CAS PubMed.
K. D. Park, J. H. Lee, S. H. Kim, T. H. Kang, J. S. Moon and S. U. Kim, Bioorg. Med. Chem. Lett., 2006, 16, 3913 CrossRef CAS PubMed.
K. Iwasa, M. Kamigauchi, M. Ueki and M. Taniguchi, European J. Med. Chem., 1996, 31, 469 CrossRef CAS.
(a) J. B. Bremner and S. Samosorn, Aust. J. Chem., 2003, 56, 871 CrossRef CAS; (b) I. I. Severina, M. S. Muntyan, K. Lewis and V. P. Skulachev, IUBMB Life, 2001, 52, 321 CrossRef CAS PubMed.
V. Milano, Y. Piao, T. LaFortune and J. de Groot, Mol. Cancer Ther., 2009, 8, 394 CrossRef CAS PubMed.
C.-Y. LO, L.-C. HSU, M.-S. CHEN, Y.-J. LIN, L.-G. CHEN, C.-D. KUO and J.-Y. WU, Bioorg. Med. Chem. Lett., 2013, 23, 305 CrossRef CAS PubMed.
M. Viale, C. Cordazzo, B. Cosimelli, D. de Totero, P. Castagnola, C. Aiello, E. Severi, G. Petrillo, M. Cianfriglia and D. Spinelli, J. Med. Chem., 2009, 52, 259 CrossRef CAS PubMed.
F. Wang, M. A. Ogasawara and P. Huang, Mol. Aspects Med., 2010, 31, 75 CrossRef CAS PubMed.

Footnotes

† Electronic supplementary information (ESI) available: Experiments and characterisation of compounds are in the ESI. See DOI: 10.1039/c4md00264d

‡ These authors contributed equally.