Preparation and optimization of a molecularly imprinted polymers – solid phase extraction system for the extraction of bioactive sesquiterpene lactones of Ambrosia maritima plant

Abstract A solid phase extraction (SPE) system for sesquiterpene lactones of damsissa was developed utilising molecularly imprinted polymers (MIPs). The prepared MIPs had a mesoporous structure and particle size of ≈2.65 µm with 3.99 nm pore size. Additionally, MIPs exhibited high thermal stability with degradation temperature between 209 and 459 °C. Optimized MIP-SPE protocol conditions were set at loading step: 1 mL ethanol; washing step: 1 mL water; eluting step: 4 mL methanol. Developed MIP-SPE system showed a binding capacity of 66.66 mg/g based on Langmuir isotherm which was selected as the best fitting model isotherm. Good selectivity coefficients were observed for neoambrosin of 2.37, 1.31 and 1.14 against umbelliferone, quercetin glucoside and p-coumaric acid, respectively. Furthermore, the proposed MIP-SPE protocol displayed some potential in the isolation of sesquiterpene lactones from damsissa plant extract and laid a foundation for the development of more selective MIPs to nonpolar natural products. Graphical Abstract


Introduction
Isolation of pure bioactive compounds form complex plant extract matrix is challenging and demanding (Peng et al. 2011). Large-scale isolation of natural products can be so costly and harmful to the environment in case of the lack of a selective adsorption media. In this scenario, huge quantities of energy and organic solvents are needed. In the recent years, molecular imprinted polymers (MIPs) became a state of art technique for extraction and isolation of pure bioactive compounds from complex matrices in a sustainable and ecofriendly manner (Vasapollo et al. 2011;Cheong et al. 2013;Ersoy et al. 2016). MIPs as synthetic material provide artificial recognition sites that can bind specifically to a target molecule if present in a mixture of structurally related compounds. So far, MIPs were extensively studied for separation of phenolics, polar diterpenes and polar triterpenoids (Bucar et al. 2013;Karimi Baker and Sardari 2021). Heretofore, sesquiterpene lactones were not tested as possible template for MIP.
Neoambrosin, ambrosin and damsin, are the major bioactive sesquiterpene lactones of the medicinal plant, Ambrosia maritima, traditionally known as damsissa ( Figure S1). These sesquiterpene lactone possess potent anticancer, anti-inflammatory and antimicrobial activities in addition to impeding the progression of rheumatoid arthritis and dementia (Abdelgaleil et al. 2011;Villagomez et al. 2013;Saeed et al. 2015;Khalil et al. 2019;Abdel-dayem et al. 2021;Mohamed et al. 2022). To date, large-scale production of these compounds is hampered by their cumbersome chemical synthesis and costly isolation procedures (Raks et al. 2018).
Thus, the current work aimed to establish an extraction scheme for sesquiterpene lactones utilizing MIPs packed in solid phase extraction (SPE) columns. The optimized MIP-SPE protocol would then be evaluated to isolate these sesquiterpene lactones from the crude alcoholic and chloroform extracts of damsissa as a trial for real sample applications. Neoambrosin was utilised as the template molecule for MIP production because of its high yield in the plant and its potent anticancer activity (Saeed et al. 2015). Methacryalic acid (MAA) and methacrylamide (MA) were selected as model monomers because of their wide applicability, availability and presence of many Hbond donor sites. Chloroform was used as a porogen because of the good solubility of template and monomers in it and its nonpolar nature which do not disrupt the interaction between the template and monomers. The fabricated MIPs were characterized and their selectivity towards sesquiterpene lactones against other structurally related compounds was evaluated.

Results and discussion
Five MIP formulae were prepared using a non-covalent imprinting technique with determined stoichiometric ratios as illustrated in Table S1. The ratio of template: monomer: cross linker was kept at 1:4:20 while varying monomers ratios (Attallah et al. 2018a(Attallah et al. , 2018b. Preliminary binding experiments were performed to determine the best prepared MIP for extraction of neoambrosin. As shown in Table S2, MIP_2 showed the highest binding capacity for neoabmrosin (55 mg/g) and was selected for further characterization.
Several techniques were applied to physically characterize the produced MIP. The scanning electron microscope (SEM) imaging of the synthesized polymers indicated that the non-molecularly imprinted polymer (NIP_2) had nonporous smooth surface, while MIP_2 formed porous microspheres ( Figure S2). The average particle sizes of MIP_2 and NIP_2 were in the range of 2416-2878 nm and 463.5-566.6 nm, respectively ( Figure S3). Such difference in the particle size could be attributed to the incorporation of the template molecule during the fabrication of MIP which led to the formation of larger particles during polymerization. This conformed to other studies where the MIPs showed bigger particle size than that of NIPs (Attallah et al. 2018a(Attallah et al. , 2018bTil et al. 2020;Saad et al. 2021). The accessible surface area and pore structure were explored utilising solid-gas isotherm profiles using Brunauer-Emmett-Teller (BET) model. It was shown that MIP_2 had lower accessible surface area (161.39 m 2 /g $ 70% lower) and lower pore volume (0.32 cc/g $80% lower) than those of NIP_2 (Table  S3). This could be attributed to the roughness heterogeneity of the NIP_2 surface due to absence of template molecule. Mesoporous structure was deduced from the low pore size ( 50 nm). The mesoporosity was also confirmed by investigating N 2 adsorption/desorption isotherm as shown in Figure S4. The isotherm adopted type IV profile according to IUPAC. This type was characterized by layer-by-layer adsorption on a smooth nonporous surface which was manifested in steep adsorption at the high relative pressure region (>0.9 P/P 0 ) (Cychosz and Thommes 2018;Til et al. 2020;Wang et al. 2020). Additionally, the type H4 hysteresis loop indicated the presence of aggregates of particles with slit-like pores in the mesoporous structures (Til et al. 2020). Spectra of FT-IR were recorded in range of 4000-400 cm À1 as illustrated in Figure  S5(a). Successful polymerization of both MIP_2 and NIP_2 was deduced from absence of absorption bands in ranges 2000-1900 cm À1 (corresponding to C ¼ C stretching) and in ranges 1600-1678 cm À1 . Peaks at $ 1719 cm À1 represented C ¼ O stretching. Strong absorption band at $ 1148 and $ 1248 cm À1 were characteristic for the symmetric and asymmetric stretching vibrations of C-O. Weak absorption bands at $ 2946 cm À1 represented the stretching vibration of C-H. Peaks at $ 1455 cm À1 indicated -CH 2bending vibration. Based on the TGA results, both MIP_2 and NIP_2 exhibited high thermal stability ( Figure S5(b)). MIP_2 and NIP_2 degradation was in range 209-459 C, 183-465 C with weight loss of 91 and 92% due to carbon skeleton degradation, respectively. The polymers showed minor weight losses (4-5%) at 93-99 C, which could be attributed to the loss of water molecules. Results were comparable to reported MIPs that were based on the same monomer utilized in the current work (Til et al. 2020).
For optimization of extraction procedures, different solvents and elution volumes were tried separately for the loading, washing and elution steps of damsissa sesquiterpene lactones. To further evaluate adsorption performance of the MIP-SPE column while optimizing these steps, maximum adsorption capacity was determined. As shown in Table S4, it turned out that EtOH was the best loading solvent, water to be the best washing solvent while MeOH was the best eluent.
Binding capacity results obtained upon extraction of neoambrosin (10-50 mg/mL) with 50 mg MIP_2 were fitted to the isotherm equations of Langmuir (Equation (S1)) (Langmuir 1918) and Freundlich (Equation (S2)) (Badruddoza et al. 2010). The best fitting for the experimental data was attained by the Langmuir model based on the obtained coefficient of determination (R 2 ). As shown in Table S5 and Figure S6, the Langmuir model for MIP_2 had R 2 of 0.9981 while R 2 was calculated to be 0.9840 for the Freundlich model. Thus, the maximum binding capacity was determined based on the Langmuir model and was found to be 66.66 mg/g. Selectivity of MIP_2 was estimated against three compounds, namely, umbelliferone, quercetin glucoside and p-coumaric acid ( Figure S7). These phenolic compounds are of common occurrence in plant extracts. Three different parameters were evaluated to determine the selectivity performance of MIPs and NIPs for neoambrosin and its structurally related compounds (Table S6). The partition coefficient (K d ) value of neoambrosin adsorbed on the MIPs (27.3 mL/g) was the highest. The obtained imprinting factor (IF) values also indicated the enhanced extent of adsorption affinity and selectivity of MIPs for neoambrosin compared with NIPs. Furthermore, the selectivity coefficient (SC) results of MIPs demonstrated that the MIPs had the strongest selective adsorption for neoambrosin.
Real alcoholic extract of damsissa containing a plethora of caffeoyl quinic acids, flavonoids and numerous sequiterpene lactones was used to evaluate the proposed MIP-SPE protocol (Salib and Michael 2007;Abdelgaleil et al. 2011;Karar et al. 2016). The HPLC chromatogram of the eluate post extraction showed a chemical profile similar to that of the alcoholic extract but with lower intensity ( Figure S8). The relative concentrations of the extracted sesquiterpene lactones to other compounds were nearly the same as the original mixture recording a recovery of 35% for damsin and neoambrosin. The application of the chloroform fraction of the alcoholic extract because of its richness in sesquiterpene lactones was also tested. However, no improvement was detected in terms of selectivity as shown in Figure S9. The limited performance of the synthesized MIP could be attributed to the overwhelming nonspecific interaction of polar compounds, rich in H-bond acceptor/donors and charged moieties, to the MIP matrix and active sites. It would be in a fashion similar to the inhibitory action of many phenolics against human enzymes (Gonc¸alves and Romano 2017). More studies are needed to improve the selectivity of MIP to non-polar template molecules in presence of high concentration of more polar vicinity, more polar concomitants. The present study was the first one to challenge MIP selectivity against real extracts rich in phenolics, chlorophyll and other small polar molecules.
Based on the reported literature, application of MIP on real samples for extraction of sesquiterpene lactones was not studied extensively. For instance, Yin et al. (2014) studied the correlation between the molecular structure of sesquiterpene lactones and the selective adsorption of MIPs and showed that the similarity of neoambrosin in structure to the 3-ring nucleus of dehydrocostus lactone promoted the selectivity of the MIPs due to stable conformation of their ring system. They also stated that the presence of minor differences in structure between template and target molecules leads to poor selectivity. Another study aimed to isolate artemisinin as a model sesquiterpene lactone utilizing MIPs based on silica gel surface showed some success in selective extraction of the desired compounds but no testing against real samples nor extracts was performed (Gong and Cao 2011).

Conclusion
Despite the advances in MIPs for extraction and isolation of natural products from plant extracts most of these natural products were phenolics, or polar alkaloids (Gong and Cao 2011;Yin et al. 2014). To the best of our knowledge, this is the first trial to isolate nonpolar compounds of the sesquiterpene lactone nucleus. Successfully, a MIP-SPE protocol for selective binding to neoambrosin was developed and optimized; this selectivity together with acceptable binding capacity was maintained in presence of other molecules having more functional group upon applying the off-line approach. However, the proposed protocol had a less satisfactory performance with real samples application, which could be attributed to the great nonspecific binding between phenolics and MIP matrix. Further improvement could be performed by using different monomers or different template. A future possible approach is to change to a more polar template such as parthenin or hymenin (a hydroxylated sesquiterpene lactone) for MIP production. Still, the presented results demonstrated a significant step forward to apply MIPs in extraction of nonpolar compounds from crude extracts.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This project received funding from the Egyptian Science, Technology and Innovation Funding Authority (STDF) Youth/one-year grant under grant agreement No. 43014.