Molecularly imprinted solid phase extraction in an efficient analytical protocol for indole-3-methanol determination in artificial gastric juice

Abstract

In this study we presented a new and efficient analytical protocol for determination of indole-3-methanol (indole-3-carbinol, I3C) in spiked artificial gastric juice used as a model sample. The analytical protocol involves separation of I3C on the imprinted stationary phase as well as the analysis of I3C using high performance liquid chromatography coupled with UV detection. The described analytical protocol allows selective isolation of I3C present in a complex matrix (such as inorganic salts and protein) from the mixture of I3C metabolites formed in artificial gastric juice with high total recovery equal to 96 ± 3% in a low concentration range of 0.125–10 μmol L⁻¹. The limit of detection and limit of quantification were 0.150 and 0.454 μmol L⁻¹, respectively. The efficacy of the analytical protocol was demonstrated by the results obtained for non-imprinted commercial sorbents C18, Florisil, and MCX Oasis® used in the separation step. The affinity of the imprinted polymer matrix was tested by non-competitive binding experiments of four structurally related analytes. The template behavior was analyzed during the imprinting process.

---

1. Introduction

Indole-3-methanol (indole-3-carbinol, I3C) represents a group of 3-substituted indoles produced from thioglucosinolate glucobrassicin present in plants belonging to the Cruciferae family such as cabbage, broccoli, cauliflower, brussels sprout, kale, turnips or rutabaga. After the consumption, enzyme myrosidase which is released while chewing, hydrolyzes thioglucoside to biologically active products such as I3C.¹ I3C was recognized as a relevant anticancer compound but its pharmacological activity is driven by multiple mechanisms which are only partly known.^2–4

It is still unclear if I3C as well as other dietary indoles act as a single factor or through complex mechanisms involving various metabolites of I3C.^5–10 These metabolites are formed at low gastric pH from I3C during the cascade of reactions and then are moved forward with chyme to other parts of the digestive system. The impact of gastrointestinal microbiota on the characteristics of subsequent fragments of the digestive system can also be significant. Slight changes in gastrointestinal pH can promote the formation of various I3C metabolites in completely different proportions.¹¹ Therefore, the benefit of oral intake of I3C could also be attributed to its metabolites formed in the gastrointestinal tract.

It is interesting to study the levels of I3C in various parts of the digestive system, especially in stomach, because such information could help to understand the transformations of I3C. In this endeavor the models of various parts of the digestive system can be very useful.¹² Such in vitro models are often applied to monitor the behavior of bioactive compounds (drugs or dietary supplements) to identify the transformation routes of those compounds and to recognize the properties of metabolites.¹³ In some cases the formed metabolites can cause fatal consequences as in the case of a combined treatment of patients with an antiviral drug, sorivudine and an anticancer agent, 5-fluorouracil.¹⁴ In the gastrointestinal tract sorivudine underwent an intensive transformation to 5-(2-bromovinyl)uracil which inactivated a liver enzyme responsible for regulation of the level of 5-fluorouracil causing its high toxic concentration. Thus, the complete knowledge about the stability of bioactive compounds in the gastrointestinal tract allows one to characterize their biological effects in vivo.

The monitoring of I3C levels in the presence of its transformation products in the digestive system required selective analytical methods. Several analytical methods described the determination of I3C after absorption in plasma or urine but none of them was applicable to the analysis of I3C in the gastrointestinal tract prior to absorption. Those methods include liquid–liquid extraction (LLE) coupled with HPLC-UV^15–17 and solid phase extraction (SPE) coupled with LC-MS/MS.¹⁸ The pharmacokinetics of I3C and its condensation products was examined by Anderton and co-workers¹⁷ in various mice peripheral tissues and brain after oral administration. It was found that the absorption, distribution and elimination of I3C occurred rapidly falling below detectable levels within one hour after intake. The level of I3C and its metabolites was quantified by Hauder and co-workers¹⁸ in plasma and urine of healthy humans after intake of glucobrassicin-rich vegetables. The concentration of I3C in plasma was below the limit of detection, but indole-3-carboxaldehyde was detected assuming fast conversion of I3C. The presented methods suffer from low recoveries and high limits of detection for I3C mainly due to an insufficient selectivity of the I3C extraction in the sample preparation step. Recently, I3C and their metabolites were analyzed in dietary supplements and plant materials. Fibigr and co-workers¹⁹ used HPLC coupled with UV to the quantitative analysis I3C of nutraceuticals. Proposed C18 sorbent allowed to isolate I3C from the commercial products but in two nutraceuticals I3C was not found. It can be result of low quality of products but it can also be a function of the sorbent properties. Phonchai and co-workers²⁰ also proposed the method for determination of indole derivatives as potential anticancer substances in dietary supplements using micellar electrokinetic chromatography with UV detection. Indole-3-methanol was analyzed simultaneously with selected indole derivatives. Pilipczuk and co-workers²¹ presented a very interesting work about determination of indolic compounds in plant extracts using HPLC-DAD-FLC detection. Sample preparation was inventive but very laborious. Indolic compounds were not isolated, but were identified in the mixture. Only compounds for which commercial standards are available were analyzed.

The development of novel analytical methods based on highly specific separation materials could be a versatile approach to dissolve many problems during the analysis of I3C. The molecularly imprinted solid phase extraction (MISPE),²² where the molecularly imprinted polymer (MIP) produced by imprinting methodology provides the stationary phase that has the desired selectivity²³ is a promising choice in selective separation techniques. The literature survey revealed only a few advanced separation materials devoted to I3C. Scorrano and co-workers²⁴ described the synthesis of MIP for the determination of I3C. The specificity of MIP was high but the total recovery of I3C (equal to 95% for MIP) was shown only for the analysis of standard samples. The utility of the method in complex matrices and in the presence of the condensation metabolites was omitted.

Although very specific methods were proposed for determination of mixture of indolic compounds, there is no efficient strategy which would enable the isolation and quantitative analysis of indole-3-methanol in the complex sample of artificial gastric juice especially in the presence of indole-3-carboxaldehyde, the main metabolite of I3C.

Thus, we decided to develop an efficient sorbent to isolation of I3C from a mixture of its metabolites formed in complex matrix using the imprinting technology. Next, we proposed a new validated analytical protocol involving optimized molecularly imprinted solid phase extraction combined with high performance liquid chromatography coupled with ultra-violet detection (HPLC-UV) for the determination of I3C in the spiked artificial gastric juice as the model sample. The artificial gastric juice was selected because its composition mimics the stomach environment where I3C is delivered after the oral intake. Such a model is often used in the analysis of compounds which can be easily metabolized in the gastric environment.^25,26 The I3C spiked model sample was incubated before analysis, which allowed us to obtain a complex mixture of I3C metabolites in a very similar composition as those present in the stomach prior to absorption. The ability of the stationary phase to extract I3C in the presence of indole-3-carboxaldehyde which was its main transformation product, was emphasized. The new analytical protocol was validated and its efficacy was demonstrated by the tests of the non-imprinted commercial sorbents C18, Florisil, and MCX Oasis® in the separation step. The affinity of the imprinted polymer matrix was analyzed by the non-competitive binding experiments of four structurally related analytes. Finally, new value was added by the testing of structural transformations of template during the imprinting process.

2. Materials and methods

2.1. Materials

Indole (IND), indole-3-methanol (I3C), 3,3′-diindolylmethane (DIM), indole-3-ethanol (I3E), indole-3-carboxylic acid (ICA), indole-3-carboxaldehyde (IAL), allylamine, polyethylene glycol dimethacrylate (P) M_w = 550 g mol⁻¹ were purchased from Sigma-Aldrich (Steinheim, Germany), ethylene glycol dimethacrylate (E) was from Fluka (Steinheim, Germany), 2,2′-azobis(2-methylpropionitrile, AIBN) was from Merck (Darmstadt, Germany). Acetone, acetic acid (glacial, 98%), hexane, dimethyl sulfoxide, ethyl acetate, formaldehyde, hydrochloric acid (36%) sulphuric acid (98%), methanol, sodium chloride were from Chempur (Piekary Śląskie, Poland) and carbon tetrachloride was from POCh (Gliwice, Poland). Pepsin from porcine gastric mucosa was delivered from Sigma Aldrich (Steinheim, Germany). Aluminium oxide for chromatography was from Merck (Darmstadt, Germany). Acetonitrile HPLC grade was from POCh (Gliwice, Poland). The monomers were purified prior to use by standard procedures (if necessary, vacuum distillation was carried out). All other reagents were used without further purification. Ultra-pure water delivered from a Milli-Q purification system (Millipore, France) was used to prepare the solutions.

2.1.1. Polymers. The experimental amounts of the reagents (moles, masses, and volumes) used for the preparation of different types of polymers are shown in Table 1. The MIPs were produced from allylamine (the functional monomer) and ethylene glycol dimethacrylate (E) or polyethylene glycol dimethacrylate (P) by the radical bulk polymerization in carbon tetrachloride acting as the porogen²⁷ using different templates: indole for the synthesis of _INDMIP_E, indole-3-methanol for the synthesis of _I3CMIP_E and _I3CMIP_P, 3,3′-diindolylmethane for the synthesis of MIP_E and _DIMMIP_P.^27,28 The details of the preparation of polymers and the simulated pre-polymerization system are described in Section S.1 in ESI.†

Table 1 Amounts of templates, cross-linkers and porogen used in the polymerization with allylamine monomer, 22.8, 0.4 (mg, mmol) and 7.9 mg of 2,2′-azobisisobutyronitrile initiator

Polymer	Template (mg, mmol)	Cross-linker (μL, mmol)	Porogen (μL)
INDMIPE	Indole, 11.7, 0.1	Ethylene glycol dimethacrylate, 377, 2.0	CCl4, 611
I3CMIPE	Indole-3-methanol, 14.7, 0.1	Ethylene glycol dimethacrylate, 377, 2.0	CCl4, 611
I3CMIPP		Polyethylene glycol dimethacrylate, 991, 2.0	CCl4, 1532
DIMMIPE	3,3′-Diindolylmethane, 24.6, 0.1	Ethylene glycol dimethacrylate, 377, 2.0	CCl4, 611
DIMMIPP		Polyethylene glycol dimethacrylate, 991, 2.0	CCl4, 1532

---

2.1.2. Solutions, metabolites and model sample. The details of the preparation and handling of stock solutions are presented in Section S.2 in ESI.† The standard solutions were prepared prior to use by dilution of the stock solutions with methanol–water (85 : 15 v/v) to obtain the required concentrations. The synthesis of indole cyclic trimer and indole cyclic tetramer as the metabolite standards was carried out according to the procedure described by Lucarini and co-workers.²⁹ The artificial gastric juice was prepared according to the formulation consisting of 0.03 M aq. sodium chloride, 0.084 M aq. hydrochloric acid and 0.32% (w/v) pepsin.³⁰

2.2. Instruments

The UV measurements were performed with a UV-1605PC spectrophotometer (Shimadzu, Duisburg, Germany). The calibration lines with the correlation coefficients r² > 0.988 for I3C, DIM, I3E, ICA, IAL standard solutions in methanol–water (85 : 15 v/v) were constructed as a peak area under the absorbance curve (y) versus concentration (x). For the right λ_max in nm, the limit of quantification (LOQ) and the limit of detection (LOD) in μmol L⁻¹ were as follows: I3C (278, 0.401, 0.132), DIM (282, 0.172, 0.057), I3E (281, 0.898, 0.296), ICA (280, 1.516, 0.500), IAL (298, 0.707, 0.233).

Reversed phase high performance liquid chromatography (RP-HPLC) was used for the determination of I3C and its condensation products in artificial gastric juice. The HPLC system consisted of an LC 20AT pump, a CTO 10A oven, an SPD-10A UV-Vis detector operating at λ = 220 nm and a Rheodyne 20 μL loop for injection. The chromatographic separation was performed using a Hypersil BDS C18 stainless steel column (150 mm × 4.6 mm ID, 5 μm, Thermo-Scientific, Waltham, MA, United States), preceded by a Bionacom® ultra filter column protector. The mobile phase consisted of the acetonitrile–water system (31 : 69 v/v) delivered at a flow rate of 1.0 mL min⁻¹ at 40 °C. The run time was 90 min followed by 90 min of equilibration (in the qualitative analysis of standards: cyclic indole trimer and cyclic indole tetramer, the run time was 120 min).

The scanning electron microscopy (SEM) analyses were performed at the Department of Chemistry, University of Warsaw, Poland. The surfaces of imprinted material was studied on Merlin FE-SEM (Zeiss, Oberkochen, Germany). The sample was Au/Pd sputter-coated before analyses.

2.3. Methods

2.3.1. Binding experiments. The stationary binding experiments were performed to evaluate the binding ability of MIPs and NIPs towards I3C. The 10 mL polypropylene tubes were filled with 10 mg of _INDMIP, _I3CMIP, _DIMMIP (E or P) as well as NIP_E and NIP_P. To each tube a volume of 5 mL of I3C standard solution of 2 μmol L⁻¹ in methanol–water (85 : 15 v/v) was added. The tubes were sealed and oscillated at room temperature for 1.5 h. Then the tubes were centrifuged for 15 min at 5000 rpm and the aliquots of supernatant (0.7 mL) were used to analyze the unbound amount of I3C by UV spectroscopy using the calibration line. For the isotherm analysis, the 10 mL polypropylene tubes were filled with 10 mg of _DIMMIP_P or NIP_P particles. A volume of 5 mL of different I3C standard solutions in the concentration range of 1–15 μmol L⁻¹ in methanol–water (85 : 15 v/v) were added. The tubes were treated in the same manner as described above.

The stationary binding experiments were performed for the affinity analysis of 3,3′-diindolylmethane (DIM), indole-3-ethanol (I3E), indole-3-carboxylic acid (ICA), indole-3-carboxaldehyde (IAL). The 10 mL polypropylene tubes were filled with 10 mg of _DIMMIP_P and a volume of 5 mL was added of each DIM, I3E, ICA, IAL standard solutions of 2 μmol L⁻¹ in methanol–water (85 : 15 v/v). Then the tubes were centrifuged for 15 min at 5000 rpm and the amounts of each analyte were determined in the same manner as described above with respective calibration lines for DIM, I3E, ICA, IAL. All analyses were made in triplicate.

The details of the calculation of binding capacities (B, μmol g⁻¹) distribution coefficients (K_D, L g⁻¹), and imprinting or affinity factors (IFs/AFs) are given in Section S.3 in ESI.†

2.3.2. Preparation of model sample. A model sample was composed of I3C and its condensation products in artificial gastric juice. An amount of 50 mg of I3C was transferred to a volume of 5 mL of artificial gastric juice in a glass round-bottom flask with a magnetic bar inside. The flask was placed in an oil-bath and stirred at 100 rpm at 37 °C for 2 h. At the beginning of the experiment the artificial gastric juice mixture was greenish and after 2 h it changed its color to light violet. At the end of the experiment a lot of precipitate was observed in the flask. After 2 h the artificial gastric juice mixture was treated with the following sequence: (i) neutralized with 0.04 M aq. ammonia solution to pH 7 (3 : 1 v/v), (ii) diluted with dimethyl sulfoxide (2 : 1 v/v) to dissolve solid components, (iii) filtered through 0.45 μm membranes and diluted with ultra-pure water (1 : 99 v/v). The dilution of the sample with water was carried out to prevent further decomposition of I3C during the analysis. The sample was analyzed by HPLC-UV to determine the amount of I3C, and the aliquot of 0.5 mL was loaded in the SPE procedure.

2.3.3. SPE of I3C from model sample. The SPE was carried out on a Macherey-Nagel manifold. Empty polypropylene 1 mL SPE columns secured by glass-fiber frits were filled with 20 mg of _DIMMIP_P particles or commercial sorbents: C18 (J. T. Baker, Phillipsburg, NJ, United States), Florisil (Fluka, Switzerland), and MCX Oasis® (Waters, MA, United States). The conditioning (0.5 mL, water) and the loading (0.5 mL) steps were applied on each column. The washing step (0.25 mL, solvent) was optimized using hexane, methanol, water, or acetonitrile, and hexane was chosen as the right solvent. The eluting step was made using 0.5 mL of methanol. The additional drying step was carried out (passing the stream of air through the polymer under gentle vacuum for max. 3 min) before and after washing with hexane. The flow rate of each SPE step was 0.5 mL min⁻¹. The elution fractions were collected and evaporated to dryness under a gentle stream of nitrogen. The dry residues were dissolved in 500 μL of ultra-pure water prior to HPLC-UV analysis.

2.3.4. Evaluation of analytical protocol. The analytical protocol for the determination of indole-3-methanol in spiked model sample of artificial gastric juice was validated by the determination of linearity, LOD, LOQ, intra and inter-day precision, accuracy as well as matrix effect. The linearity of calibration line for I3C was analyzed in the range of 0.125–10 μmol L⁻¹ for artificial gastric juice sample spiked with the standard solutions of I3C at six different concentrations (0.125, 0.25, 0.5, 1, 5 and 10 μmol L⁻¹). The standard solutions of I3C were added to model sample after neutralization of the artificial gastric juice with ammonia solution in order to avoid the transformation of I3C in acidic solution. The peak area ratios of I3C were plotted against the corresponding concentrations and the calibration line was constructed by means of the least-squares method. The LOD and LOQ for I3C were determined as the signal-to-noise (S/N) ratio of 3 and 10, respectively. The precision of the method of I3C analysis was evaluated using intra- and inter-day variations. The relative standard deviation (RSD) was determined as the measurement of precision. The intra- and inter-day robustness of I3C analysis were determined in triplicate within one day and over two consecutive days, respectively. The intra-day and inter-day accuracy for I3C was measured as a bias (systematic distortion). In order to examine the matrix effect, the total recoveries of I3C were analyzed in elution fractions after SPE from neutralized artificial gastric juice spiked with standard solutions of I3C at concentrations of 0.125, 0.25, 0.5, and 1 μmol L⁻¹. The total recoveries of I3C were calculated with respect to the amount of I3C present in the neutralized model sample before the loading step of SPE.

The reusability of new imprinted sorbent _DIMMIP_P was examined by three consecutive SPE processes on the same sorbent. The elution fractions were analyzed after each SPE from neutralized artificial gastric juice spiked with standard solutions of I3C at concentration of 0.125 μmol L⁻¹.

The impact of components of artificial gastric juice as well as the metabolites of I3C on I3C determination was considered. In order to identify I3C metabolites, the commercial standards of I3C, IAL, DIM were used, and the indole cyclic trimer as well as the indole cyclic tetramer were synthesized (Section 2.1.2.).

3. Results and discussion

3.1. Design of sorbent

3.1.1. Choice of template. Allylamine was selected as the most promising functional monomer for production of sorbents for I3C isolation based on the preliminary measurement of I3C binding capacity on the 3,3′-diindolylmethane imprinted polymer. Three template molecules built from the indole system were tested to form selective polymer matrices with ethylene glycol dimethacrylate (E): _INDMIP_E (imprinted by indole), _I3CMIP_E (imprinted by indole-3-methanol), _DIMMIP_E (imprinted by 3,3′-diindolylmethane). For comparison, respective NIP_E was synthesized without any template molecules. The stationary binding experiments of I3C were carried out for those polymers in order to determine the binding capacities B, the distribution coefficients K_D, and the IF/AFs (Section S.3 in ESI†).

The obtained results are presented in Table 2 in the first four lines. The lowest affinity was noted for _INDMIP_E which was synthesized in the presence of indole as the template, and the highest affinity was observed for _I3CMIP_E which was synthesized in the presence of I3C. These values are consistent with the results observed for the compound which is both an analyte and a template. However, the _DIMMIP_E polymer which was synthesized in the presence of 3,3′-diindolylmethane revealed also high binding capacity towards I3C. The lowest distribution coefficient of _INDMIP_E could be the evidence that a single indole system without any substituent is insufficient to create selective binding sites for I3C, and therefore indole was excluded from further studies.

Table 2 Binding capacities, B distribution coefficients, K_D and calculated imprinting/affinity factors, IF/AF of _INDMIP_E, _I3CMIP_E, _DIMMIP_E and NIP_E as well as _I3CMIP_P, _DIMMIP_P and NIP_P for I3C (conc. 2 μmol L⁻¹, n = 3)

Polymer	Binding capacities ± S.D. (B, μmol g^−1)	Distribution coefficient (KD, L g^−1)	IF/AF
INDMIPE	0.076 ± 0.002	0.040	1.7
I3CMIPE	0.192 ± 0.005	0.115	5.0
DIMMIPE	0.109 ± 0.003	0.063	2.7
NIPE	0.045 ± 0.001	0.023	—
I3CMIPP	0.178 ± 0.006	0.106	6.9
DIMMIPP	0.239 ± 0.008	0.153	10.1
NIPP	0.030 ± 0.001	0.015	—

---

3.1.2. Choice of cross-linker. Two remaining templates were used to prepare polymers using polyethylene glycol dimethacrylate (P) as the cross-linker. P was proposed on the basis of the results obtained in our previous paper,²⁸ where the polymer matrix formed from polyethylene glycol dimethacrylate showed very high affinity towards indole derivatives. Thus, new MIP coded as _I3CMIP_P was synthesized using I3C as the template and the second polymer coded as _DIMMIP_P was synthesized de novo²⁸ using 3,3′-diindolylmethane as the template. The respective non-imprinted NIP_P was synthesized from polyethylene glycol dimethacrylate and allylamine without any template. The _I3CMIP_P, _DIMMIP_P and NIP_P polymers were evaluated towards I3C as described in Section 2.3.1. and the results of the binding capacities, B, distribution coefficients K_D, and IFs/AFs are presented in Table 2 in the last three rows. As it could be seen, the dimethacrylate P with a longer spacer –OCH₂CH₂O– produced different polymer matrices than ethylene glycol dimethacrylate E. The non-imprinted polymer NIP_P yielded much lower binding capacity than NIP_E. Out of two imprinted polymers, _I3CMIP_P showed slightly lower binding capacity than _I3CMIP_E synthesized from the ethylene glycol dimethacrylate cross-linker, but _DIMMIP_P showed much higher binding capacity. The AFs are very high: 6.9 and 10.1, respectively (a significant increase of AF is partially due to lower non-specific adsorption). Thus, we decided to select _DIMMIP_P as the promising sorbent for the purpose of this study.

3.2. Analytical protocol for determination of I3C in model sample

The artificial gastric juice was spiked with I3C prior to incubation at 37 °C, which allowed us to obtain a complex mixture of I3C metabolites with indole-3-carboxaldehyde (the main I3C transformation product) and other I3C condensation products in similar composition and proportion as those produced in the stomach. Such a complex mixture was a model sample in further investigations of the new analytical protocol. The mixture was used in the optimization process of SPE using _DIMMIP_P as the sorbent. The new analytical protocol for the determination of I3C in the presence of its metabolites consisted of analyte separation on _DIMMIP_P and the analysis of eluent using HPLC-UV.

3.2.1. Analysis of I3C and its condensation products in spiked artificial gastric juice as a model sample. After the incubation, the sample was neutralized and treated with dimethyl sulfoxide in order to dissolve all components of the sample prior to HPLC-UV analysis (Section 2.3.2.). The obtained chromatogram of the model sample is presented in Fig. 1a. As it can be seen, numerous peaks are present in the chromatogram which belongs to I3C and its metabolites. Based on commercial standards, the peak of I3C was identified at t_r = 2.88 min (Fig. 1b), the peak of its derivative, indole-3-carboxaldehyde (IAL) was identified at t_r = 3.43 min. However, to identify the peaks of I3C condensation products (other than 3,3′-diindolylmethane), we had to synthesize indole cyclic trimer and indole cyclic tetramer (Section 2.1.2.). The isolation of these compounds was very difficult but the HPLC analysis allowed us to identify their peaks at the following retention times: indole cyclic trimer at t_r = 73.75 and indole cyclic tetramer at t_r = 98.71 min (the chromatogram of a fraction of products after the first purification step is presented in Fig. S1 in ESI†). We deduced that both high intensity peaks at t_r = 35.00 min and t_r = 65.30 min shown in Fig. 1a could not belong to indole cyclic trimer or to indole cyclic tetramer. Thus, the findings of Anderton and co-workers¹⁷ or Larson-Su and Williams³¹ were considered to assign the remaining peaks. We can suppose that the peaks at t_r = 35.00 and at t_r = 65.30 min belong to indole linear trimer and to indole linear tetramer, respectively.

Fig. 1 Chromatograms of: (a) the model sample of artificial gastric juice spiked with I3C after 2 h at 37 °C (inside square box a 90 min range chromatogram is shown), (b) the standard of I3C, (c) the elution fraction after SPE of I3C on _DIMMIP_P of the model sample. The retention time of: indole-3-carbinol is 2.88, indole-3-carboxaldehyde is 3.43 min. The assignment of peaks of indole linear trimer – 35.00, and indole linear tetramer – 65.30 min based on literature data.^17,31

As a result of the above findings we concluded that in the model sample of artificial gastric juice, I3C was oxidized to indole-3-carboxaldehyde as well as condensed into linear oligomers of high molecular weight but 3,3′-diindolylmethane DIM was practically not detected. This finding is in agreement with the investigations of Hauder and co-workers¹⁸ which revealed that indole-3-carboxaldehyde and indole-3-carboxylic acid were found in plasma as the only metabolites of I3C.

The amount of I3C after 2 h of incubation was equal to 14.1 ± 0.4 mg comprising 28.2 ± 0.8% of the initial amount of I3C used at the beginning of the experiment.

3.2.2. SPE procedure on _DIMMIP_P. The new SPE protocol was proposed for separation of I3C from the model sample of artificial gastric juice on _DIMMIP_P. The model sample preparation was finished by the addition of dimethyl sulfoxide to dissolve solid components and dilute the sample, thus the loading solution consisted of water and dimethyl sulfoxide. It was necessary to optimize diligently the washing step in order to remove impurities as well as non-specifically bound compounds but not to remove the analyte of I3C. The following solvents were selected for the washing optimization: methanol, water, acetonitrile, and hexane. The criterion of choice based on the physico-chemical properties of solvents: protic–polar (methanol and water), aprotic–polar (acetonitrile) and aprotic–non-polar (hexane). The washing step was carried out separately for each solvent prior to the elution with methanol. The percentage of I3C in eluates (the recovery) was determined by HPLC-UV analysis. The recoveries were presented as the percentage of the amount of I3C adsorbed on the sorbent after the loading step and are as follows: methanol, 3.2 ± 0.2%, water, 4.8 ± 0.2%, acetonitrile, 12.7 ± 0.4%, and hexane, 98 ± 3% (n = 3). The results revealed that non-polar and aprotic solvents enhance the specific adsorption of I3C. Hence, hexane was selected as the most appropriate washing solvent. Methanol was selected as the eluent based on the previously published results.²⁴

The SPE procedure was combined with HPLC-UV analysis of I3C. The new analytical protocol revealed its superior ability for selective separation of I3C from the complex mixture of its metabolites. The chromatogram of the elution fraction after optimized SPE protocol on _DIMMIP_P and HPLC-UV analysis is presented in Fig. 1c.

The chromatogram revealed the highest intensity peak of I3C at t_r = 2.88 min and very low intensity peaks of its main metabolite, indole-3-carboxaldehyde at t_r = 3.43 min as well as low intensity peaks at t_r = 35.00 min and t_r = 65.98 min (not shown on chromatogram in Fig. 1c). This could be the evidence that selective separation of I3C was achieved, and their determination is possible.

3.2.3. Validation of analytical protocol. The parameters such as linearity, LOQ, LOD, intra and inter-day precision, accuracy as well as matrix effect were defined for I3C in order to characterize the new analytical protocol for the determination of I3C in the model sample of artificial gastric juice using optimized SPE on _DIMMIP_P coupled with HPLC-UV analysis.

The calibration line was established for I3C to cover a low concentration range from 0.125 to 10 μmol L⁻¹. A good linearity of the calibration line (y = 17050139x − 702) was obtained with regression coefficient r² = 0.994. LOD was 0.150 μmol L⁻¹ (22 μg L⁻¹) and LOQ was 0.454 μmol L⁻¹ (67 μg L⁻¹). In order to examine the applicability of new analytical protocol, the intra and inter-day precision was analyzed. The results showed excellent repeatability and intermediate precision for determination of I3C in proposed method. In the repeatability analysis, three extractions were carried out in the same day and the RSDs values ranged between 4.3 and 5.8%. The intermediate precision was determined by measuring the RSDs of experiments performed in various days. The inter-day (day-to-day) precision for I3C was between 3.5 and 7.6%. The intra-day and inter-day accuracy for I3C measured as a bias (systematic distortion) were between −7.8 and +7.4% as well as −8.9 and +7.8%, respectively.

The effect of complexity of the model sample was considered as the effect of components of artificial gastric juice (pepsin and inorganic salts). The recoveries of I3C were analyzed for four concentrations (Section 2.3.4.) and ranged between 94.6 ± 5.9 and 101.3 ± 5.7% (n = 3). The excellent recovery obtained for the novel method showed its utility and applicability to model sample analysis. The results confirmed that the presence of salts as well as protein had no impact on recovery of I3C in the analyzed concentration range.

The reusability study were carried out on the same sorbent (in the same cartridge) in three consecutive SPE processes. The total recoveries of each sequential SPE process were as follows: 102.0 ± 9.2%, 98.1 ± 4.7%, and 96.4 ± 3.4%, respectively. The results shows that imprinted sorbent can be used minimum three times without significant lost of its separation capabilities.

3.2.4. Quantitative analysis of I3C using new validated protocol. It is well known that the inorganic salts or proteins affect the adsorption on the imprinted sorbent decreasing its binding capacity. Moreover, the I3C metabolites – the compounds that possess the indole system in their structure and are present in the model sample, could compete with I3C to the specific binding sites in the imprinted polymer network decreasing its binding capacity and selectivity.²³ The satisfactory results of analyses revealed that this method may be suitable for determination of I3C in artificial gastric juice. No significant impact of the sample complexity was shown on the recovery of I3C. The total recovery of I3C on _DIMMIP_P was equal to 96 ± 3% (n = 5).

3.2.5. Analysis of I3C using non-imprinted commercial sorbents in SPE step. The new analytical protocol was also carried out using three commercial sorbents in the separation step: C18 (non-polar sorbent), Florisil (polar sorbent), and MCX Oasis® (ion-exchange sorbent) to compare their usefulness with the imprinted sorbent _DIMMIP_P to the I3C isolation. The total recoveries of I3C from the model sample (artificial gastric juice spiked by I3C) were as follows: for C18, 31 ± 3%, for Florisil, 50 ± 3%, and for MCX Oasis®, 9 ± 2% (n = 5). The total recoveries on commercial sorbents were not good enough, and the isolation of I3C was not efficient. The results showed the superiority of the _DIMMIP_P imprinted sorbent to separation of I3C from the complex matrix.

3.3. Characterization of sorbent

The characterization of _DIMMIP_P was provided in the paper²⁸ in which the composition of matrix (by ¹³C CP/MAS NMR spectroscopy), the particle surface (by scanning electron microscopy and Brunauer–Emmett–Teller isotherm) were discussed. Additionally, the micrographs of _DIMMIP_P in Fig. 2 well documented the repeatability of synthetic procedure of polymer matrix.

Fig. 2 Micrographs of _DIMMIP_P.

3.3.1. Adsorption isotherm of I3C on _DIMMIP_P. The I3C adsorption on _DIMMIP_P was examined employing the Freundlich model defined by the equation: B = aF^m where B is the bound amount of the analyte and F is the unbound amount of the analyte, a is the measure of capacity (B_max) and m is a heterogeneity index. The Freundlich model fitted well MIP adsorption data in the low concentration regions and allowed us to determine surface heterogeneity of the tested material. The plot of I3C adsorption in double logarithmic scale is presented in Fig. S2 in ESI.† The straight line (log B = 0.109 log F − 0.943, r² = 0.949) is the evidence that the adsorption of I3C can be described by the Freundlich equation. The estimated value of m was 0.11, which showed that the imprinting process provided very heterogeneous population of binding sites in _DIMMIP_P matrix (the heterogeneity increased as the value of m decreased).

3.3.2. Affinity of _DIMMIP_P towards structurally related analytes. Non-competitive dynamic binding experiments were performed to analyze the recognition property and selectivity of _DIMMIP_P. A group of structurally related compounds was selected to perform the test: 3,3′-diindolylmethane (DIM), indole-3-ethanol (I3E), indole-3-carboxylic acid (ICA), indole-3-carboxaldehyde (IAL). The results were compared with the indole-3-methanol (I3C) binding value (the chemical formulas of the tested compounds are presented in Fig. 3).

Fig. 3 The chemical formulas of the tested compounds.

The low concentration of 2 μmol L⁻¹ of the standard solution of each compound was applied to be sure that selective adsorption of the analyzed compounds was preferred. The details of the calculation of binding capacities, B and distribution coefficients, K_D for DIM, I3E, ICA, IAL and I3C are presented in Section S.3 in ESI.† The results are shown in Table 3. The highest binding capacities on _DIMMIP_P were observed for I3C. Its interactions with _DIMMIP_P were strong and selective. The binding capacity of DIM was also high, which is logical because DIM was used as the template during polymerization. The remaining analytes were characterized by much lower binding capacities, which can be the evidence that the lack of the methylene linker or lengthening them to the ethylene group make the interaction with the polymer network hard to create.

Table 3 Binding capacities, B and distribution coefficients, K_D of the tested analytes on _DIMMIP_P (conc. 2 μmol L⁻¹, n = 3)

Compound	Binding capacity ± S.D. (B, μmol g^−1)	Distribution coefficient (KD, L g^−1)
I3C	0.239 ± 0.008	0.153
DIM	0.188 ± 0.003	0.099
I3E	0.118 ± 0.002	0.047
ICA	0.102 ± 0.006	0.064
IAL	0.067 ± 0.001	0.041

---

Two of the tested compounds: indole-3-carboxylic acid (ICA) and indole-3-carboxaldehyde (IAL) are the products of I3C transformation and IAL is the main I3C metabolite found in the model sample (spiked I3C artificial gastric juice) (Section 3.2.1.) as well as in real samples.¹⁸ Their peaks appeared on the chromatogram close to the peak of I3C at low retention times. Therefore, a significant difference in distribution coefficients, K_D between I3C and ICA or IAL (0.153 to 0.064 or 0.041 L g⁻¹, respectively) is an additional merit in the analysis based on the molecularly imprinted sorbent _DIMMIP_P.

3.3.3. Template behavior during imprinting process. The behavior of the template of DIM was analyzed during the imprinting process. At the beginning of sorbent preparation (during the removal of the template from polymer matrix), we observed many peaks in the control chromatograms of the fractions collected during DIM template removal (see Fig. S3 in ESI†) which could be assigned to the indole derivatives. We were interested in finding out when such transformation happened. We supposed that DIM transformation started in the pre-polymerization system. The _DIMMIP_P pre-polymerization system was simulated by the solution containing the template (DIM), the porogen (carbon tetrachloride) and the initiator of free radical polymerization (AIBN), but the monomers were omitted (Section S.2 in ESI†). The mixture was purged with nitrogen, and heated for one hour at 64 °C (the temperature of initiator decomposition). Then, the composition of the mixtures was examined by HPLC-UV. The chromatogram of the so-called ‘false’ pre-polymerization solutions F_DIM is presented in Fig. 4.

Fig. 4 A chromatogram of the so-called ‘false’ polymerization mixture for F_DIM. The retention time of: indole-3-carbinol is 2.84, indole-3-carboxyaldehyde is 3.34, and 3,3′-diindolylmethane is 9.13 min.

As it can be seen, numerous peaks are present in the chromatogram together with the peak at t_r = 9.13 min which was identified as the template, 3,3′-diindolylmethane. Based on standards, the following peaks were assigned to: I3C at t_r = 2.84 min and IAL at t_r = 3.34 min. It could be concluded that in the tested pre-polymerization system which at the beginning contained 3,3′-diindolylmethane, carbon tetrachloride and the initiator, the complex mixtures of compounds were formed and subsequently imprinted into the polymeric network as the templates.

It can be supposed that three factors might affect the conversion of 3,3′-diindolylmethane to I3C: the presence of a free radical initiator, the elevated temperature, and the presence of trace amounts of intrinsic water. In our opinion, the most convincing explanation can be based on the studies of Błoch-Mechkour and co-workers^32,33 who investigated radical cations of IND, I3C and DIM generated by radiolysis in ionic liquid or Ar matrices at cryogenic temperature. It could be expected that a similar process of free radical cascades occurs in the pre-polymerization mixture of F_DIM leading to the transformation of the 3,3′-diindolylmethane molecule. A similar chromatographic pattern was observed for _DIMMIP_E. The I3C molecules present in the polymerization mixture can be imprinted into the _DIMMIP_P as well as _DIMMIP_E polymer matrices.

Finally, the significantly higher binding capacity of I3C on _DIMMIP_P than on _DIMMIP_E (AF = 10.1 and 2.7, respectively, see Table 2) should be commented on. Here, the hydrophilic property of the cross-linker can play a crucial role. The unusual property of poly(ethylene oxide) and its unexpected water solubility and hydrophilicity was discussed by Israelachvili.³⁴ It could be expected that a more hydrophilic network formed from polyethylene glycol dimethacrylate could favor the interaction with I3C from the water–methanol standard solutions. The polymeric network formed from ethylene glycol dimethacrylate has less hydrophilic character.

4. Conclusions

In this study we developed the newly optimized analytical protocol employing HPLC-UV with the polymeric network of 3,3-diindolylmethane imprinted poly(allylamine-co-polyethylene glycol dimethacrylate) _DIMMIP_P as the efficient sorbent in the MISPE step of indole-3-methanol (I3C) determination. The advantage of this procedure is the ability of the stationary phase to extract I3C in the presence of indole-3-carboxaldehyde which was its main transformation product. The efficient analytical protocol showed a high recovery value of I3C (96 ± 3%) from the complex model sample of artificial gastric juice which was impossible to obtain using the C18, Florisil, and MCX Oasis® commercial sorbents. The transformation analysis of the template molecule, 3,3′-diindolylmethane can be crucial in the interpretation of imprinting process of polymer matrix with indolic compounds.

References

1. J. V. Higdon, B. Delage, D. E. Wiliams and R. H. Dashwood, Pharmacol. Res., 2007, 55, 224–236.
2. Y. S. Kim and J. A. Milner, J. Nutr. Biochem., 2005, 16, 65–73.
3. G. L. Firestone and S. N. Sundar, Mol. Endocrinol., 2009, 23, 1940–1947.
4. B. B. Aggarwal and H. Ichikawa, Cell Cycle, 2005, 4, 1201–1215.
5. S. Safe, S. Papineni and S. Chintharlapalli, Cancer Lett., 2008, 269, 326–338.
6. G. Brandi, M. Paiardini, B. Cervasi, C. Fiorucci, P. Filippone, C. De Marco, N. Zaffaroni and M. Magnani, Cancer Res., 2003, 63, 4028–4036.
7. J. E. Riby, C. Feng, Y.-C. Chang, C. M. Schaldach, G. L. Firestone and L. F. Bjeldanes, Biochemistry, 2000, 39, 910–918.
8. L. Xue, C. M. Schaldach, T. Janosik, J. Bergman and L. F. Bjeldanes, Chem.–Biol. Interact., 2005, 152, 119–129.
9. Y.-C. Chang, J. Riby, G. H. F. Chang, B. C. Peng, G. Firestone and L. F. Bjeldanes, Biochem. Pharmacol., 1999, 58, 825–834.
10. M. De Santi, L. Galluzzi, S. Lucarini, M. F. Paoletti, A. Fraternale, A. Duranti, C. De Marco, M. Fanelli, N. Zaffaroni, G. Brandi and M. Magnani, Breast Cancer Res., 2011, 13, R33.
11. K. R. Grose and L. F. Bjeldanes, Chem. Res. Toxicol., 1992, 5, 188–193.
12. M. Neumann, K. Goderska, K. Grajek and W. Grajek, Żywność. Nauka. Technologia. Jakość, 2006, 1, 30–45 , in Polish.
13. T. Sousa, R. Paterson, V. Moore, A. Carlsson, B. Abrahamsson and A. W. Basit, Int. J. Pharm., 2008, 363, 1–25.
14. H. Okuda, K. Ogura, A. Kato, H. Takubo and T. Wanabe, J. Pharmacol. Exp. Ther., 1998, 287, 791–799.
15. J. Moussata, Z. Wang and J. Wang, J. Chromatogr. B: Anal. Technol. Biomed. Life Sci., 2014, 958, 1–9.
16. M. J. Anderton, M. M. Manson, R. D. Verschoyle, A. Gescher, J. H. Lamb, P. B. Farmer, W. P. Steward and M. L. Williams, Clin. Cancer Res., 2004, 10, 5233–5241.
17. M. J. Anderton, R. Jukes, J. H. Lamb, M. M. Manson, A. Gescher, W. P. Steward and M. L. Williams, J. Chromatogr. B: Anal. Technol. Biomed. Life Sci., 2003, 787, 281–291.
18. J. Hauder, S. Winkler, A. Bub, C. F. Rufer, M. Pignitter and V. Somoza, J. Agric. Food Chem., 2011, 59, 8047–8057.
19. J. Fibigr, D. Satinsky, L. Havlikova and P. Solich, J. Pharm. Biomed. Anal., 2016, 120, 383–390.
20. A. Phonchai, P. Wilairat and R. Chantiwas, Anal. Methods, 2016, 8, 637–643.
21. T. Pilipczuk, N. Dawidowska, B. Kusznierewicz, J. Namieśnik and A. Bartoszek, Food Anal. Methods, 2015, 8, 2169–2177.
22. D. Djozan, M. A. Farajzadeh, S. M. Sorouraddin and T. Baheri, J. Chromatogr. A, 2012, 1248, 24–31.
23. C. Alvarez-Lorenzo and A. Concheiro, Handbook of Molecularly Imprinted Polymers, Smithers Rapra Technology, Shawbury, 2013.
24. S. Scorrano, L. Mergola, R. Del Sole, M. R. Lazzoi and G. Vasapollo, J. Appl. Polym. Sci., 2014, 131, 40819.
25. B. Muszyńska, K. Kała, K. Sułkowska-Ziaja, A. Krakowska and W. Opoka, Food Chem., 2016, 199, 509–515.
26. Y. S. Lee, J.-K. Moon, K.-H. Liu, E. Kim, H. Choi and J.-H. Kim, J. Korean Soc. Appl. Biol. Chem., 2014, 57, 397–405.
27. P. Luliński, D. Klejn and D. Maciejewska, Polym. Int., 2014, 63, 695–702.
28. D. Klejn, P. Luliński and D. Maciejewska, Mater. Sci. Eng., C, 2015, 56, 233–240.
29. S. Lucarini, M. De Santi, F. Antonietti, G. Brandi, G. Diamantini, A. Fraternale, M. F. Paoletti, A. Tontini, M. Magnani and A. Duranti, Molecules, 2010, 15, 4085–4093.
30. A. B. Stefaniak, M. A. Virji, C. J. Harvey, D. C. Sbarra, G. A. Day and M. D. Hoover, Int. J. Hyg. Environ. Health, 2010, 213, 107–115.
31. S. Larson-Su and D. E. Williams, Toxicol. Sci., 2001, 64, 162–168.
32. A. Błoch-Mechkour, T. Bally, A. Sikora, R. Michalski, A. Marcinek and J. Gębicki, J. Phys. Chem. A, 2010, 114, 6787–6794.
33. A. Błoch-Mechkour, T. Bally and A. Marcinek, J. Phys. Chem. A, 2011, 115, 7700–7708.
34. J. Israelachvili, Proc. Natl. Acad. Sci. U. S. A., 1997, 94, 8378–8379.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra23593j

---