Magnetic COFs as Satised Support for Lipase Immobilization and Recovery to Effectively Achieve the Production of Biodiesel by Great Maintenance of Enzyme Activity

Background: Production of biodiesel from renewable sources such as inedible vegetable oils by enzymatic catalysis has been a hotspot but remains a challenge on the ecient use of an enzyme. COFs (Covalent Organic Frameworks) with large surface area and porosity can be applied as ideal support to avoid aggregation of lipase and methanol. However, the naturally low density limits its application. In this work, we reported a facile synthesis of core-shell magnetic COF composite (Fe 3 O 4 @COF-OMe) to immobilize RML (Rhizomucor miehei lipase), to achieve its utilization in biodiesel production. Result: This strategy gives extrinsic magnetic property, and the magnetic COFs is much heavier and could disperse in water medium well, facilitating the attachment with the enzyme. The resultant biocomposite exhibited an excellent capacity of RML due to its high surface area and fast response to the external magnetic eld, as well as good chemical stability. The core-shell magnetic COF-OMe structure not only achieved highly ecient immobilization and recovery processes but also maintained the activity of lipase to a great extent. RML@Fe 3 O 4 @COF-OMe performed well in practical applications, while free lipase did not. The biocomposite successfully achieved the production of biodiesel from Jatropha curcas Oil with a yield of about 70% in the optimized conditions. Conclusion: Magnetic COFs (Fe 3 O 4 @COF-OMe) for RML immobilization greatly improved catalytic performance in template reaction and biodiesel preparation. The magneticity makes it easily recovered and separated from the system. This rst successful attempt of COFs-based immobilized enzyme broadened the prospect of biodiesel production by COFs with some inspiration. and facilitated the separation taking advantage of features of high surface area, porosity, good chemical stability, and strong magnetic response. The support could maintain the activity of RML and perform well in practical applications. We applied the immobilized RML into the production of biodiesel from un-editable Jatropha curcas Oil and obtained a satisfactory yield. This rst successful attempt of COFs-based immobilized lipase demonstrates their promising applications in the production of biodiesel. The N 2 sorption isotherms measured at 77 using a liquid N 2 bath. Scanning electron microscopy (SEM) images collected using a JSM-5900LV (JEOL, Japan). Transmission electron (TEM) images collected using a H-600, operating at 75kV. IR spectra were recorded on a Nicolet Impact 410 FTIR spectrometer. The confocal laser scanning microscopy (CLSM) data were collected on a Leica SP5 under an excitation λ ex = 488 nm and λ ex =543 nm. The vibrating sample magnetometer (VSM) data were measured by LakeShore7404. Thermal Gravity analysis (TG/TGA) was tested and analyzed by METTLER TOLEDO TGA/DSC2/1600.

Preparation and characterization of Fe 3 O 4 @COF-OMe nanoparticles and immobilized RML.
The facile synthesis of magnetic core-shell COFs is based on the room-temperature synthesis of COF-OMe (Fig S1-S4). The detailed preparation and immobilization process is illustrated in Fig.1, which involved two main steps: (1) coprecipitation synthesis of magnetic Fe 3 O 4 nanoparticles and rapid roomtemperature synthesis of the core-shell structured magnetic Fe 3 O 4 @COF-OMe composites in a one-pot process by mixing Fe 3 O 4 nanoparticles (30 mg, 0.13 mmol) as the magnetic core and 2,5-dimethoxyterephthalaldehyde (DMTP, 0.24 mmol) and 1,3,5-tris(4-aminophenyl)-benzene (TPB, 0.16 mmol) as building units of COF-OMe in the acetonitrile according to the result of morphology (Fig. S5). (2) immobilization process of RML by physical absorption in PBS buffer.
The as-prepared biocomposites could be applied in the production of biodiesel.
The morphologies of COF-OMe, Fe 3 O 4 @COF-OMe, and RML@Fe 3 O 4 @COF-OMe are veri ed by scanning electron microscopy (SEM) and transmission electron microscopy (TEM), shown in Figure 2. It can be seen that COF-OMe has good dispersity and exhibits a uniform nanosphere structure (Fig. 2(A)) with a size of 500-600 nm. The core-shell structure of Fe 3 O 4 @COF-OMe with the thickness of the COF shell about 70 nm is con rmed by the TEM image ( Fig. 2(B)). The Fe 3 O 4 @COF-OMe particles display similar spherical morphology as COF-OMe ( Fig. 2(C)). And after enzyme immobilization, the composite morphology remains unchanged, only the surface becomes rough (Fig. 2(D)).
Fourier transforms infrared (FT-IR) spectroscopy is carried out to prove that the successful synthesis of Fe 3 O 4 @COF-OMe. As shown in Figure S6, The FTIR spectrum of Fe 3 O 4 contains a band at 579 cm -1 , which is assigned to characteristic Fe-O-Fe stretch. The characteristic absorption bands at 1610cm -1 assigned to the C=N stretch mode observed in the curve of Fe 3 O 4 @COF-OMe means the successful synthesis of COF-OMe by condensation of aldehydes and amines.
Along with the characteristic Fe-O-Fe stretch found in the curve of biocomposite, the combination of COF-shell and magnetic Fe 3 O 4 was proved, demonstrating the successful preparation of Fe 3 O 4 @COF-OMe. 41 The crystalline structure of Fe 3 O 4 @COF-OMe is examined by PXRD patterns (Fig. 3(A)). XRD image exhibits 6 peaks with 2θ at 30 29 This is found in the PXRD image of COF-OMe pattern, shown inset. The peak at 2.75° is attributed to the plane (100) of COF-OMe. The other planes, like (110), (200), (210), (220) corresponds to peaks at 4.8°, 5.5°, 7.3° and 9.2°. 45 This pattern con rms the formation of the crystalline form of COF-OMe. These successful and facile preparation represents it an alternative way of traditional synthesis of them, which provides guidance for the exploration of other COFs.
Thermogravimetry analysis (TGA) expounds on the thermal stability and different components of biocomposites, shown in Fig. 3(B) and Fig. S7 (DTG). For COF-OMe, there is a distinct decrease in weight that occurs at 300-400 ℃, which means its structure begins to disintegrate. In other words, a long plateau under 419 o C demonstrates the high thermal stability of COF-OMe. As for RML@Fe 3 O 4 @COF-OMe, the weight loss at about 280 o C can be attributed to the removal of lipase. The two parts of mass losses occurred at 48 o C and 410 o C is consistent with it of bare Fe 3 O 4 (4 % at 48 o C) and COF-OMe (12 % at 410 o C) respectively. The sharp weight-loss at over 700 o C may due to the reaction between melt COF-OMe and Fe 3 O 4 of core-shell structure. In a word, the support Fe 3 O 4 @COF-OMe displays such satisfactory thermal stability as COF-OMe, where the TG curve runs smoothly under 400 o C. At the same time, the core-shell structure doesn't react mutually under 700 o C, which means it is quali ed to be a good carrier of an enzyme.
The magnetic property of these nanospheres is characterized by a vibrating sample magnetometer (VSM). The magnetic hysteresis curve ( Fig. 3(C)) of nanomagnetic Fe 3 O 4 has an excellent magnetic property, with a saturated magnetization value of 46.07 emu g -1 . There is a drop observed in Fe 3 O 4 @COF-OMe (~20 emu g -1 ) and RML@Fe 3 O 4 @COF-OMe (~6 emu g -1 ), which are attributed to the loading of COF shell and enzyme. Despite this, rapid aggregation of biocomposites from the suspension is obtained with the help of an external magnet, which could reduce the desorption of the enzyme by centrifugation in this way.
Nitrogen sorption isotherms measured at 77 K indicates the BET surface of Fe 3 O 4 @COF-OMe decreases from 232 cm 2 g -1 to 28 cm 2 g -1 after immobilization of RML ( Fig. 3(D)). The pore-size distribution analyses of Fe 3 O 4 @COF-OMe and lipase@Fe 3 O 4 @COF-OMe calculated by the density functional theory have shown that both of the samples have a pore size centered at about 3.1 nm, whereas the pore volume drops from 0.223 cc g -1 to 0.036 cc g -1 after RML absorption ( Figure S8 & Table S1). The result indicates the successfully loading of lipase, and it suggests that the magnetic COFs may serve as a promising carrier for lipase immobilization.
To further verify the distribution of RML on the support, the Fluorescein-labelled enzyme is an optical way to prove its existence and determine its distribution.
Fluorescent probe uorescein isothiocyanate (FITC) is used to label the enzyme molecules (green) generally. 44,46 However, it is not available to use in RML@COF-OMe in this work. This is because the support, COF-OMe itself, is uorescent. Under an excitation λ=488 nm (the parameter of FITC-labelled protein), the long emission at λ=490-690 nm is got by the COF-OMe itself, which interferes with the detection of the FITC-labelled enzyme, so it is incapable to prove the existence of enzyme on the surface of the carrier and determine its distribution ( Figure S9). In this case, Rhodamine B isothiocyanate (RBITC)labeled RML was prepared. The RBITC-labelled RML (red) is present throughout Fe 3 O 4 @COF-OMe (green), which is observed by CLSM analysis at excitation wavelengths of 488 nm for Fe 3 O 4 @COF-OMe and 543 nm for RBITC-RML, demonstrating that the enzyme accommodated in this composite ( Figure S10).
The optimization of the immobilization process time, concentration of lipase, and temperature were studied during the immobilization process. The enzyme loading of Fe 3 O 4 @COF-OMe increased with the time at the beginning ( Figure 4(A)), and decreased after 8h. It can be explained that long-term shaking caused leakage of lipase after absorption saturation. The temperature made an in uence on both immobilization e ciency and activity of RML. In the immobilization process, we mixed the support and lipase in 4 o C and 25 o C respectively. We found there was an improvement in RML attaching with stirring at room temperature ( Figure 4(B)). The lipase solution with the initial concentration of 10, 20, 40, 80 mg/L was prepared. Though the relative immobilization e ciency decreased as the ratio of the enzyme increased, the total amount of immobilized RML further increased with a higher concentration of RML ( Figure 4(C)). According to the results, the immobilization process was undergoing at 25 o C, mixing 80mg/L of lipase with 10mg support for 8 hours.
After the immobilization, the hydrolysis of p-NPA (see details in Support information) was adapted to examine whether the immobilized RML is active and its stability. In this work, we employed the core-shell magnetic COFs (Fe 3 O 4 @COF-OMe) to enhance the recovery e ciency. Here, we also compared this strategy with the common mixing method. This tactic is to make COFs magnetic by mixing COFs and magnetic As shown in Fig. 5. After immobilization, there was a shuttle decrease in activity in hydrolysis of p-NPA of both immobilized RML (Fig. 5(A)). The best outcome could recover to 60% of the free enzyme (Fe 3 O 4 @COF-OMe) as the time prolonged. The Fe 3 O 4 @COF-OMe also showed good thermal and pH stability. The stability of the activity for both free RML and immobilized enzyme in different pH ranging from 5.0 to 10.0 was studied and plotted in Fig. 5(B). The result showed that the optimal pH altered slightly, from about 7.0 to 8.0. Thermal stability was investigated, which the biocomposites were stored at 60 ℃ for 12h ahead of tests. It was observed that there is a decrease in activity for all of them, but the range of decrease was not signi cant for Fe-COFs immobilized RML ( Fig. 5(C)). We found that although the Fe 3 O 4 -COF-OMe has a higher RML uptake, the enzyme activity of it did not perform well. It is due to the non-uniform and solid two-phase framework prepared by physical mixing and adhesion strategy, which was not enough to maintain the RML activity.
Based on the outcomes, Fe 3 O 4 @COF-OMe with better thermal stability and activity can indeed be optimal support for the subsequent study of transesteri cation reaction.

Activity Assay of transesteri cation reaction
The activities of both free lipase and immobilized RML were determined by a transesteri cation reaction between 2-phenyl ethanol and vinyl acetate ( Figure  S13). First of all, both Fe 3 O 4 @COF-OMe and Fe 3 O 4 -COF-OMe were adapt to catalyze under the same conditions. We found that this result was consistent with it of hydrolysis of p-NPA (Fig. 5) where it was Fe 3 O 4 @COF-OMe that performed better than Fe 3 O 4 -COF-OMe (Fig. 6(A)). At the same time, in the transesteri cation reaction, the yields of both immobilized enzymes were higher than that of free enzymes, which proved the excellent protective effect of carriers on the enzyme.
The solvent effect of the reaction in which n-Hexane functioned as a solvent was shown in Fig. 6(B), with the highest yield up to 80%. Interestingly, according to the results, we found that the yields altered by the trends of the polarity of different solvents. In detail, the more hydrophobic the solvent was, the higher yields we got. As for the optimized solvent, whose log P value was largest, the yield was much higher than the others at the same time. Carbon tetrachloride, trichloroethylene, and toluene, whose polarity was similar, had almost the same yields of 20%. However, if the solvent was hydrophile, such as THF, acetone, the transesteri cation didn't happen in it. To furtherly verify the hydrophobic solvents were conducive to this reaction, several homologous liquids of n-Hexane were adapted (Fig. S14). n-Hexane, c-Hexane, and n-Heptane had similar yields, which indicated the hydrophobic solvents were bene cial in this work.
Then we investigated the in uence of temperature on the reaction (Fig. 6(C)). Immobilized RML did better in the transesteri cation than the free enzyme. With the rise of temperature, the yield of the immobilized enzyme increased gradually and reached a peak at about 50 ℃, while of liquid enzyme decreased continuously. The loss of activity may due to the conformation change of RML caused by high temperature, which affected the binding of the active center and substrate. It demonstrated that the carrier could effectively protect the enzyme from heat and kept its catalytic activity.
The dosage of RML has also played an important role in a transesteri cation reaction, where excessive enzyme not only causes waste but also reduces the rate due to the aggregation. So here, we studied the yields of the reaction with different amounts of RML. As we can see, in Fig. 6(D), the yield of free RML still went up along with the increase of amount. For Immobilized RML, the yield didn't show a signi cant rise when the RML on carrier changed to 0.8 mg.
Considering the e ciency and economy of this reaction, 0.5 mg RML was used in every single sample assay. At the optimal conditions, n-Hexane as a solvent, the transesteri cation yield can reach about 80% with 0.5 mg of immobilized RML at 50 ℃.
The preservation of activity by the protection of support in organic solvents and high temperature were shown above, where the immobilized RML always did better in different organic solvents and at over 30 ℃ than free lipase. To furtherly assess the function of COFs in protecting the catalytic ability of RML, the tolerance of immobilized RML against ultrasonic operation was investigated. As shown in Figure S15, the yields of immobilized RML did not change signi cantly after ultrasonic treatment, and always higher than that of free RML at the same time. Here we also studied the leakage ratio of RML by washing the immobilized lipase (Fig. 7). As we can see, the amount of loss of RML for every single wash was about 2% and the total leakage ratio was less than onefth after the 8 wash cycle, which indicated a good ability to preserve the lipase from washing operation.

Production of biodiesel
Having established the e ciency of RML@Fe 3 O 4 @COF-OMe in the transesteri cation reaction of 2-phenol ethanol and vinyl acetate, then we studied its catalytic ability in the production of biodiesel from inedible Jatropha curcas Oil ( Table 1). The outcomes a catalyzed by immobilized RML are much better than those by free RML, with a yield of 67.8% and 5.1% respectively. It is noticed that there is an obvious loss in enzymatic activity when the amount of methanol exceeds the stoichiometric ratio (3:1). This is due to the inhibitory effect of methanol, and the activity is irreversibly inactivated. 36 Compared to free RML, magnetic Fe 3 O 4 nanoparticles could protect RML from the methanol, as the product could be detected with a satisfactory yield. It is noticed that the protective effect is not permanent although the activity could be maintained if the concentration of methanol is doubled (Entry 4, Table 1). But if the amount of methanol is excessive too much (15:1 and 30:1), there is a huge loss in yield. In a word, the nanoparticle e ciently improved the stability and maintained the activity of the enzyme in practical application.

Conclusion
In summary, we successfully prepared a core-shell magnetic COFs structure (Fe 3 O 4 @COF-OMe) in a facile way. The resultant magnetic COFs exhibited a great uptake of RML and facilitated the separation taking advantage of features of high surface area, porosity, good chemical stability, and strong magnetic response. The support could maintain the activity of RML and perform well in practical applications. We applied the immobilized RML into the production of biodiesel from un-editable Jatropha curcas Oil and obtained a satisfactory yield. This rst successful attempt of COFs-based immobilized lipase demonstrates their promising applications in the production of biodiesel.

Materials
All chemicals and reagents were commercially available and used without further puri cation. 2,5-dimethoxyterephthalaldehyde (DMTP), 1,3,5-tris(4- Characterization Powder X-ray diffraction (PXRD) data were collected on X'Pert Pro MPD DY129 X-ray diffractometer (40 kV, 40 mA) using Cu Kα (λ = 1.5406 Å) radiation. The gas adsorption isotherms were collected on a surface area analyzer, Quantachrome Instruments version 3.01 Autosorb Station 2. The N 2 sorption isotherms were measured at 77 K using a liquid N 2 bath. Scanning electron microscopy (SEM) images were collected using a JSM-5900LV (JEOL, Japan). Transmission electron microscopy (TEM) images were collected using a HITACHI H-600, operating at 75kV. IR spectra were recorded on a Nicolet Impact 410 FTIR spectrometer. The confocal laser scanning microscopy (CLSM) data were collected on a Leica SP5 under an excitation λ ex = 488 nm and λ ex =543 nm. Ca=(C 0 -C) V/M s C 0 represents the initial concentration of RML solutions, while C is the concentration of supernatant after immobilization. V is the total volume of the system. M s is the mass of support.
Enzymatic activity after immobilization process was measured through the hydrolysis of p-NPA. To a tube, p-NPA (500 μL, 2 μmol/mL) and PBS buffer (2 mL, pH=7.4) with immobilized RML (or 2 mL total volume of free enzyme and PBS buffer solution) were added. The absorbance at 405 nm was detected 1h later.
Finally, the product concentrations were corrected for the auto-hydrolysis of p-NPA and also the absorbance of p-NPA left in the solution. (See details in Support Information). Ahead of thermal stability tests, both free RML and Immobilized RML were stored at 60 o C for 12h.
Activity assay for transesteri cation reaction.
The transesteri cation of 2-phenylethanol and vinyl acetate were chosen to determine the activity of immobilized RML in production of biodiesel. Generally, 2phenylethanol (20 μL) and vinyl acetate (40 μL) were added into a tube with 2 mL of solvent, 150rpm for 24h. Different bio-complex, solvent, temperature, and amount of RML in carrier were optimized. The yields of transesteri cation were determined by HPLC analysis with a Waters-HPLC on C18 columns using methanol/water (70:30) as the eluent.

Production of Biodiesel
In the experiment of production of Biodiesel, to a tube of 3 mL of the solvent containing Jatropha curcas Oil (0.15 mmol) and methanol (0.45 mmol), immobilized RML was mixed. Then the system was put into a shaker at 150rpm for 48 h at conditions optimized by transesteri cation of 2-phenylethanol and vinyl acetate. The yields were determined by GC-analysis. The synthesis of core-shell magnetic nanoparticles and immobilization process.