Verbascoside: Identification, Quantification, and Potential Sensitization of Colorectal Cancer Cells to 5-FU by Targeting PI3K/AKT Pathway

Colorectal cancer (CRC) is the third most common cancer mortality worldwide. Although, 5-Fluorouracil (5-FU)-based chemotherapeutic regimens remain the mainstay for treatment of CRC, intrinsic and acquired resistance to 5-FU is the main reason for treatment failure and relapse. Adjunct or add-on therapy, therefore, should be thought of to enhance responsiveness to 5-FU. Verbascoside (VER) is a phenylethanoid glycoside ingredient present in many Plantago species and was widely used in traditional medicine. VER showed antiproliferative effects in many cancer types including CRC. In the present study, VER in Plantago seeds was identified using UPLC-MS/MS and quantified using newly developed and validated UPLC-DAD followed by investigating its potential sensitization of CRC cells to 5-FU in vitro. The potential impact on PI3K/AKT pathway was also investigated. A synergistic cytotoxic interaction between 5-FU and VER besides G1 cell cycle arrest were detected. Enhanced apoptosis mainly by affecting Bax and Bcl-2 and to a lesser extent Bcl-xL and p53 was also observed. Additionally, 5-FU combined to VER was capable of significantly reducing PI3K and p-AKT/total AKT ratio. Overall, these results suggest a potential role of VER as an adjuvant treatment to decrease the resistance of CRC cells to 5-FU possibly by targeting the PI3K/AKT pathway.

Plant extraction. Fifty grams of seeds of P. pysillium L. and P. afra L., each finely grinded and extracted with 80% hydromethanol. The hydromethanolic extract was evaporated at reduced pressure with rotary evaporator (BÜCHI, Germany) at 45 °C and kept in refrigerator till further analyses.

Identification of verbascoside using UPLC-ESI-MS/MS.
For identification of VER in P. pysillium L. and P. afra L. using UPLC-MS/MS system, the optimum conditions were as follows: The mobile phase consisted of the same two solvents; (A): Deionized water containing 0.1% formic acid and (B): Methanol containing 0.1% formic acid. The flow rate was 0.4 ml/min and the gradient time (t G ) was 24 min: isocratic with 10% B for 0-0.6 min, after gradient from 10% B to 90% B from 0.6 to 20 min, afterwards gradient 90% B to 100% B from 21-22 min and lastly the initial conditions 10% B from 22-24 min. The parameters for analysis were carried out using negative ion mode as follows: source temperature 150 °C. Desolvation gas temperature 400 °C, cone voltage 30 eV, capillary voltage 3 kV. The flow rate of Nitrogen gas as cone gas and desolvation gas was 50 and 900 L/h, respectively. Mass spectra were detected in the ESI negative ion mode between m/z 100-900. The peaks and spectra were processed using the Maslynx 4.1 software and tentatively identified by comparing its retention time and mass spectrum with reported data.
Identification and quantification of verbascoside using UPLC-DAD. For separation and quantification of VER in P. pysillium L. and P. afra L. using UPLC-DAD system, the optimum conditions were as follows: The mobile phase consisted of two solvents; (A) Deionized water containing 0.1% formic acid and (B) Methanol containing 0.1% formic acid. The gradient time (t G ) was 20 min: Isocratic with 10% B for 0-0.6 min, after gradient from 10% B to 50% B from 0.6 to 6 min, then isocratic with 50% B from 6 to 11 min, afterwards gradient 50% B to 90% B from 11-13 min and kept for 2 minutes, finally back to the initial conditions 10% B from 15-17 min and remained at this condition for 3 min. The flow rate was 0.4 ml/min, the temperature was 35 °C, the injection SCiEnTiFiC RepoRts | (2018) 8:16939 | DOI:10.1038/s41598-018-35083-2 volume was 10 μl and the UV detector was adjusted at 332 nm. It is worth mentioning that the mobile phase solvents; in both systems, were filtered using 0.22 μm filter and then degased for 15 min before utilization.
The identification of the VER; in both plant extracts; was carried out by comparing the retention time with its standard. While the quantification of VER was accomplished utilizing the established calibration curve.
The standard and the plant extract preparation. One mg/ml stock of VER standard was prepared by dissolving 10 mg VER in 10 ml MeOH. Working solution (0.1 mg/ml) was prepared by dilution from the stock solution using MeOH. Ten mg of either P. pysillium L. or P. afra L. was dissolved in 10 ml methanol then ultrasonicated for 15 min and filtered with 0.22 μm syringe filter.
Method validation. The method was validated according to the ICH guidelines 20 and the following parameters were investigated: Linearity was evaluated by constructing a calibration curve; in which the area was plotted versus the concentration of the VER corresponding standards. In order to establish the calibration curve, 10 VER standards (6,8,10,20,40,60,80,100,200, and 400 μg/ml) were prepared by serial dilution from the stock solution. Each one of these standards was injected three times on the system. The limit of quantification is the concentration at which standard deviation of the intercept/slope ratio was equal to 10 and the limit of detection is the concentration at which standard deviation of the intercept/slope ratio was equal to 3.3. Repeatability of the method was assessed by injecting three concentrations (10, 20, and 40 μg/ml) for 3 times on the same day and then relative standard deviation (% RSD) was calculated. Meanwhile, intermediate precision was evaluated by injecting the same three concentrations for three consecutive days and the % RSD was determined. Accuracy was evaluated by comparing the actual found concentration by the system to the theoretical value of ten standards of the calibration curve. The results were expressed in terms of percentage recovery (%R) and % RSD. Further investigation for the accuracy took place by spiking 1 ml of 0.1 mg/ml of each plant extract with known concentration of standard (10, 20, and 30 μg/ml). Each sample; for each plant; was injected three times then the % recovery and the % RSD were calculated.  21 . Briefly, cells were seeded in 96-well plates and cultured overnight. The following day, cells were treated with the above concentrations of 5-FU and/or VER as well as DMSO. Seventy-two hours later, 20 μl of MTT was added to each well followed by a 2 h incubation period after which the absorbance was measured. Cell viability was calculated as a percentage of the absorbance at 570 nm in treated cells against that of control. All experiments were performed in triplicates. Half-maximal inhibitory concentration (IC 50 ) was calculated using GraphPad Prism software, version 5.00 (GraphPad Software, CA, USA). For studying the nature of drug interaction, the combination index (CI) was calculated according to the following formula 22 : pair 50 50 pair 50 The interaction was described as synergistic if CI < 1; antagonistic if CI > 1; and additive if CI equals to 1. (Bio-Rad, CA, USA) was used for cDNA synthesis and PCR amplification steps, according to the manufacturers' instructions, as described previously 23 . The relative mRNA expression levels [fold change from untreated control samples normalized to glyceraldehyde phosphate dehydrogenase (GAPDH) as the housekeeping gene] of p53, Bcl-2 (B-cell lymphoma 2), Bax (Bcl-2 associated X protein), and Bcl-xL (B-cell lymphoma-extra large) were assessed using the 2 −ΔΔCt analysis method, as previously described by Livak and Schmittgen 24 . The primer sequences used for RT-qPCR are listed in Table 1.

Apoptosis assay using Annexin
Determination of caspase-3, -8, and -9 activity. In order to determine the potential of treatments to induce apoptosis and investigate the underlying mechanism of action, levels of active caspase-3, caspase-8, and caspase-9 were measured using human caspase-3, caspase-8, and caspase-9 ELISA kits obtained from Invitrogen (CA, USA), according to the manufacturers' instructions.

Determination of total PI3K activity.
To determine the effect of 5-FU, VER, and 5-FU + VER on total PI3K activity, human PI3K ELISA kit (MyBioSource, CA, USA) was used according to the manufacturer's instructions. Absorbance was measured at 450 nm.

Determination of total AKT and phospho-AKT (Ser473).
To determine the effect of 5-FU, VER, and 5-FU + VER on total AKT (t-AKT), and phosphorylated AKT (p-AKT) activity; ELISA kit for the determination of t-AKT and p-AKT (Ser473) (Abcam, MA, USA) was used according to the manufacturer's instructions. Briefly, cells were seeded in 96 well-culture plates and after 24 hrs, the cells were exposed to the different treatments. 72 hrs later, media was removed and both t-AKT and p-AKT were determined in cell lysates. The p-AKT/total AKT ratio was then calculated.

Statistical analysis.
All values are presented as means ± standard deviation (S.D.) from three independent experiments performed in triplicates. Statistical analysis was performed by one-way ANOVA followed by Bonferroni post hoc test using GraphPad Prism, version 5.0 (GraphPad Software, CA, USA) (v5). Statistical significance was determined at P < 0.05.

Results and Discussion
Identification and suggested fragmentation of verbascoside. Vebascoside (VER) or acteoside, a phenylethanoid glycoside composed mainly of a sugar skeleton of β-glucose and rhamnose monosaccharaides whereas the caffeoyl and hydroxyl phenylethyl aglycons replaced the hydroxyl groups of C 4 and C 1 of β-glucose, respectively as shown in the structure in Fig. 1. Total ionization chromatogram in the negative mode (TIC) of both Plantago species revealed the presence of VER at retention times 7.68 and 7.69 min. of P. pysillium L. and P. afra, L., respectively as in Fig. 2 a,b. Identification of VER through its fragmentation was observed in the presented fragmentation as shown in Fig. 1 and the daughter ion spectrum as shown in Fig. 3 [25][26][27] . An ion peak at m/z 135 could be an evidence of anhydrophenethanol moiety 27 as shown in Fig. 1.

Method development and validation of VER on UPLC-DAD. Several methods have been investigated
to reach the optimum method. At the beginning, different solvents were investigated; water 0.1% acetic acid and MeOH, then doubling the concentration of % acetic acid was also investigated. After, another solvent was tested; water 0.1% formic acid and MeOH and finally the best resolution and least retention time were achieved using water 0.1% formic acid and MeOH 0.1% formic acid. Accordingly, it was selected as the optimum solvent. It is worth mentioning that doubling the concentration of formic acid did not have a dramatic effect on the resolution or the retention time (tR). After selecting the optimum solvent, further investigation regarding the gradient time took place, till the optimum was reached, as illustrated in the experimental part. The temperature was adjusted at 35 °C, higher temperature didn't improve the resolution, neither had a dramatic effect on the tR decrease. The wavelength was adjusted at 332 nm, as per the UV spectrum of VER determined by the diode array detector (DAD). VER standard was injected under the optimum conditions and its peak appeared at 6.793 min as in Fig. 4 (a). The method was validated according to the ICH guidelines. To evaluate the linearity of the method, a calibration curve was constructed as mentioned in the experimental part. The method was linear over the concentration range of 6-400 μg/ml, the regression equation was y = 341.34 × −0.6013. The good linearity of the method was confirmed by the value of the regression coefficient (r) 0.9999. The limit of quantification was 4 μg/ml and the limit of detection was 1.35 μg/ml.
The repeatability of the method was evaluated by injecting the three concentrations (10, 20, and 40 μg/ml), three times each on the same day and the % RSD were 0.72, 1.63, and 0.33%, respectively. For the intermediate precision, the same three concentrations were injected each three times for three consecutive days and the % RSD were 1.54, 1.80, and 1.73%, respectively. The low values of the % RSD for both repeatability and intermediate precision indicate high precision of the method.
The accuracy of the method was determined by comparing the actual concentration found by the system to the theoretical value of ten standards of the calibration curve. And the % recovery ranged from 97.72 to 101.73%, with % RSD ranging from 0.09 to 1.91%.
The newly developed and validated method was applied on the identification and quantification of VER on P. afra L. and P. pysillium L. Accordingly, each plant extract was injected separately under the optimum conditions; the peak that appeared at tR 6.780 min corresponds to VER. This was identified by peak matching with VER standard, as presented in Fig. 4.
The constructed calibration curve was employed to quantify VER in the plant extracts. Hence, it was found that P. afra L. contains 19.01 μg VER/mg of extract, while P. pysillium L. contains 25.56 μg VER/mg of extract.
Additionally, both P. pysillium L. and P. afra, L. were spiked with VER standards at 3 different levels to further ensure the accuracy of the method and the quantification of VER.
As 1 mg of P. afra was found to contain 19.01 μg VER, accordingly, it was spiked with 10, 20, and 30 μg/ml, so that the final concentration is 29.01, 39.01, and 49.01 μg, respectively. The % recoveries were 96.28%, 97.13%, and 95.54% with % RSD 0.972%, 0.814%, and 1.12%, respectively. On the other hand, one mg of P. psyllium L. was found to contain 25.56 μg VER. Hence, it was spiked with 10, 20, and 30 μg/ml, so that the final concentration is 35.56, 45.56, and 55.56 μg, respectively. The % recoveries were 99.32%, 99.06%, and 95.48% with % RSD 0.79%, 2.09%, and 1.36%, respectively.   Cytotoxicity and cell cycle analysis. MTT assay showed that 5-FU, VER, 5-FU + 0.01 µM VER, and 5-FU + 0.1 µM VER decreased the rate of cell proliferation in HCT-116 and Caco-2 cell lines, as compared to control (Fig. 5). The IC 50 of 5-FU and VER was determined and found to be 1.199 and 1.088 µM on HCT-116 cells as well as 0.269 and 0.956 µM on Caco-2 cells, respectively. Combining a low dose of VER (0.01 μM) to 5-FU was found to produce an antagonistic interaction where the CI was found to be 3.45 and 1.25 on HCT-116 and Caco-2 cell lines, respectively. However, adding a high dose of VER (0.1 μM) to 5-FU had interestingly resulted in a synergistic interaction on both cell lines where the CI was found to be 0.43 and 0.25 on HCT-116 and Caco-2 cells, respectively (Table 2). Hence, treatment of Caco-2 cells with 5-FU+0.1 μM VER was used for further investigations. Moreover, as shown in Fig. 5, cell cycle analysis performed on Caco-2 cells exposed to 5-FU and VER, either alone or combined, caused cell cycle arrest at G1 with a 12.44-, 10.27-, and 18.49-fold increase in the cells in the pre-apoptotic phase after treatment with 10 μM 5-FU, 0.1 μM VER, and 10 μM 5-FU + 0.1 μM VER, respectively. In a previous study, VER was also found to significantly inhibit the proliferation of colorectal cancer cells 28 .
Effect on caspase-3, caspase-8, and caspase-9. As shown in Fig. 6, Treatment of Caco-2 with 5-FU, VER, and 5-FU + VER caused 6.7-, 4.3-, and 7.1-fold increase in the levels of caspase-3, a crucial mediator of apoptosis, as compared to control, respectively. This shows that combining VER to 5-FU caused a 1.1-fold further increase in caspase-3 levels compared to 5-FU alone. Moreover, in order to explore the mechanism of action by which apoptosis was initiated in cancer cells, the effect on caspase-8 and -9 was further investigated. Regarding caspase-8, treatment with 5-FU, VER, and 5-FU + VER caused a 4.8-, 3.5-, and 8.8-fold increase in its level as compared to control, respectively. Accordingly, the increase in caspase-8 in the 5-FU + VER-treated group was 1.8-fold higher compared to 5-FU alone. As for caspase-9, 5-FU, VER, and 5-FU + VER showed a 9.9-, 4.4-, and 11.3-fold higher expression levels compared to control, respectively, with a 1.8-fold higher expression reported in the combination group compared to 5-FU-treated group.
In human cells, induction of apoptosis involves either an intrinsic or extrinsic pathway. These pathways are triggered and induced either through Fas (extrinsic) or Bax (intrinsic). Fas activates pro-caspase-8 29,30 which then activates caspase-8 to cleave and further activate caspase-3 and other downstream caspase enzymes. Cytochrome-c is then released by the mitochondria under the influence of the pro-apoptotic Bax 31 . Pro-caspase-9, the initiator of the intrinsic apoptosis pathway, then becomes activated by cytochrome-c, consequently activating caspase-3 beside other caspases 32,33 . Hence, the results of the present study might suggest the involvement of both intrinsic and extrinsic pathways. Further insights into the apoptotic mechanism of action will be discussed later.
Effect on Bax, Bcl-2, BclxL, and p53 gene expression. For further assessment of the apoptotic potential of the combination treatment, the gene expression of Bax, Bcl-2, Bcl-xL, and p53 was estimated in Caco-2 cells treated with 5-FU, VER, alone and combined (Fig. 6). Regarding Bax gene expression, the present results showed that treatment with 5-FU, VER, and 5-FU + VER caused a 14.48-, 2.46-and 32.59-fold higher expression than control untreated cells, respectively, where the combination-treated cells expression was 2.25-fold higher than that in 5-FU-treated cells. Our results also showed that Bcl-2 expression was decreased in 5-FU-, VER-, and 5-FU + VER-treated cells by 74, 91, and 89.33%, as compared to control untreated cells, respectively. Additionally, the combination treatment showed a 58.96% decrease in Bcl-2 expression, as compared to 5-FU alone. On the other hand, the expression of Bcl-xL showed no significant difference in the combination treated group compared to control in spite the fact that both 5-FU and VER alone showed 1.26-and 1.74-fold increase in the expression compared to control, respectively. Moreover, the levels of p53 was almost 2-fold higher in all treated groups (with no significant differences between them) compared to control. These results are in accordance with the study of   Zhou et al. 28 and strongly suggest a potential apoptotic effect of VER in colorectal cells especially when combined with 5-FU. Thus, it can be concluded that the apoptotic potential of VER, especially when combined with 5-FU, could be primarily due to its effect on Bax and Bcl-2 rather than Bcl-xL and p53. Effect on PI3k inhibition. As shown in Fig. 7, VER showed no change in PI3K level compared to control, however, cells treated with 5-FU caused a 83.18% reduction in PI3K compared to untreated cells. Moreover, combining VER to 5-FU caused a further significant reduction in PI3K reaching 89.14 and 35.59%, as compared to control and 5-FU, respectively. There is no evidence in the literature for the effect of VER on PI3K/AKT pathway except for the study of Wang et al. 34 , which reported an upregulation of PI3K by VER in neurons suggesting a neuroprotective role against beta-amyloid-induced neurotoxicity, however, no reports for an influence on tumor cells, in general, or CRC cells, in particular were found. To our knowledge, the present study offers the first evidence for a potential effect of VER, particularly when combined with 5-FU, on PI3K/AKT pathway in CRC. Although, VER alone failed to inhibit PI3K contrary to 5-FU, it unexpectedly caused a significant downregulation in its level when combined to 5-FU. These findings need further studies to explore the exact mechanism by which VER potentiated the effect of 5-FU on PI3K when used in combination.
Effect on total AKT, p-AKT, and p-AKT/total AKT ratio. Total AKT was reduced in the cells treated with 5-FU and 5-FU + VER, however, the VER-treated cells showed higher AKT expression compared to control. Interestingly, different pattern was observed for p-AKT where the expression in 5-FU-treated cells was the same as that of control while VER-treated cells showed higher expression than that in untreated control cells. The combination, however, showed the lowest levels of p-AKT as compared to control, 5-FU-, and VER-treated cells. In order to provide a more clear picture of the observed results, the p-AKT/total AKT ratio was calculated where 5-FU-and VER-treated cells showed 1.13-and 1.36-fold higher ratio compared to control, however, the combination treatment showed more than 1 fold lower ratio than that of the control untreated cells (Fig. 7). Although, 5-FU remains a common treatment for patients with colon cancer, resistance to the drug was widely reported leading to therapeutic failure 35 . Resistance of CRC cells was previously linked to the upregulation of PI3K/AKT pathway in these cells and hence, suppressing this pathway has been suggested for sensitizing cancer cells to conventional treatment 36,37 . These findings are in agreement with our results where 5-FU caused the p-AKT/ total AKT ratio to increase significantly compared to control implying resistance of CRC to 5-FU. In spite the fact that VER, when used alone, showed a similar pattern to that of 5-FU, the combination showed a different pattern where the p-AKT/total AKT ratio was significantly decreased compared to untreated cells. As mentioned earlier regarding the pattern observed with PI3K, further studies are recommended in order to unravel the mechanism by which VER synergized with 5-FU to decrease the p-AKT/total AKT ratio and hence cause an overall downregulation of the PI3K/AKT pathway leading to sensitization of CRC cells to 5-FU. In conclusion, the present findings suggest a potential role of VER in reducing the resistance of CRC to 5-FU via targeting the PI3K/AKT signaling pathway. Further studies, however, are recommended to validate the in vivo efficacy of the combination.

Data Availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.