Assessing Enzyme Immobilization on Reverse Asymmetric Membranes and Biocatalytic Reactor Performance

23 Integration of membrane filtration and biocatalysis has appealing benefits in terms of simultaneous 24 substrate conversion and product separation in one reactor. Nevertheless, the interaction between 25 enzymes and membrane is complex and the mechanism of enzyme docking on membrane is similar to 26 membrane fouling. In this study, focus is given on the assessment of enzyme immobilization mechanism 27 on reverse asymmetric polymer membrane based on the permeate flux data during the procedure. 28 Evaluation of membrane performance in terms of its permeability, fouling mechanisms, enzyme 29 loading, enzyme reusability and biocatalytic productivity were also conducted. Alcohol Dehydrogenase 30 (EC 1.1.1.1), able to catalyze formaldehyde to methanol with subsequent oxidation of NADH to NAD 31 was selected as the model enzyme. Two commercial, asymmetric, flat sheet polymer membranes (PES 32 and PVDF) were immobilized with the enzyme in the reverse mode. Combination of concentration 33 polarization phenomenon and pressure driven filtration successfully immobilized almost 100% of the 34 enzymes in the feed solutions. The biocatalytic membrane reactor recorded more than 90% conversion, 35 stable permeate flux with no enzyme leaching even after 5 cycles. The technique showing promising 36 results to be expanded to continuous membrane separation setup for repeated use of enzymes. 37 38 39 40 41 42 43


Abstract 23
Integration of membrane filtration and biocatalysis has appealing benefits in terms of simultaneous 24 substrate conversion and product separation in one reactor. Nevertheless, the interaction between 25 enzymes and membrane is complex and the mechanism of enzyme docking on membrane is similar to 26 membrane fouling. In this study, focus is given on the assessment of enzyme immobilization mechanism 27 on reverse asymmetric polymer membrane based on the permeate flux data during the procedure. 28 Evaluation of membrane performance in terms of its permeability, fouling mechanisms, enzyme 29 loading, enzyme reusability and biocatalytic productivity were also conducted. Alcohol Dehydrogenase 30 (EC 1.1.1.1), able to catalyze formaldehyde to methanol with subsequent oxidation of NADH to NAD 31 was selected as the model enzyme. Two commercial, asymmetric, flat sheet polymer membranes (PES 32 and PVDF) were immobilized with the enzyme in the reverse mode. Combination of concentration 33 polarization phenomenon and pressure driven filtration successfully immobilized almost 100% of the 34 enzymes in the feed solutions. The biocatalytic membrane reactor recorded more than 90% conversion, 35 stable permeate flux with no enzyme leaching even after 5 cycles. The membrane was cut and fixed, reverse mode in a 50 ml dead-end Amicon stirred cell (Milipore, 122 USA). Reverse mode is when the support layer is facing the feed and the skin layer is supported by 123 another polypropylene layer. The membrane was soaked in a mixture of ethanol (50% v/v) for 10 124 minutes following the manufacturer's instruction. The cleaned membrane will be filtered with ultrapure 125 water at a pressure of 3 bar and 100 rpm in stirred cell in normal orientation (skin layer facing feed). 126 After compression, 50 ml of ultrapure water is introduced in the stirred cell and the membrane is set up 127 in reverse mode to determine the permeability with applied pressure of 1 bar and 100 rpm of stirring. 128 129

Enzyme immobilization 130
The feed for immobilize reaction contained 0.1 g/L of ADH in a 100 mM phosphate buffer solution pH

Contact angle measurement 145
The surface hydrophilicity of PES & PVDF membrane is measured by using contact angle goniometer 146 instrument (AST/VCA-3000s). The angle of the water and both membranes is measured. The 147 measurement is taken 3 times for both membranes and the results are compared. 148 149

Determination of various parameters 150
The percentage of enzyme loading is the calculation to determine the efficiency of enzyme 151 immobilizing on/in the membrane. The formula of the percentage of enzyme loading is: 152 where is the amount of the enzyme that immobilized and is the amount of enzyme in the feed 154

solution. 155
Flux recovery ratio (FRR) is a calculation to identify the type of fouling resistance that forms on the 156 membrane surface by taking the flux of the permeate. The formula for FRR is: 157 (2) 158 Total fouling ratio (Rt), reversible resistance (Rr) and irreversible resistance (Rir) are calculated as 159 follow: 160 (3) 161 (4) 162 Where = permeate flux; 1 = initial pure water flux and 2 = final pure water flux 164 The conversion rate of reaction is calculated to determine the conversion of NADH by the enzyme. 165 8 The formula of the conversion rate is: 166 where is the concentration of NADH in the feed solution and is the concentration of NADH in 168 permeate solution. 169 Biocatalytic productivity is a calculation to calculate the efficiency of the enzyme converting substrate 170 to product. The formula for the biocatalytic productivity is: 171 where is the mass of the production and is the mass of the enzyme.  During the filtration of ADH, a high local concentration of enzyme will arise near the support layer-197 solution interface, due to a balance between the convective drag force towards and through the 198 membrane and back transport away from the membrane (Figure 1). This phenomenon is called 199 concentration polarization (CP). CP will result in flux decline due to the increment of the osmotic 200 pressure near the surface of the membrane. Hence, the effectiveness of pressure-driven filtration is 201 reduced. Figure 3 shows the flux profile during enzyme immobilization which shows 3 phases of flux 202 decline related to enzyme deposition mechanism in/on the membrane. The initial phase (Phase 1) 203 includes macromolecular sorption and particle deposition. Fresh membrane is exposed to the ADH and 204 adsorb onto the polypropylene fibrous strand (Figure 1c&d). Part of the ADH may aggregate because 205 of electrostatic forces (from ionizable chemical species in the buffer) and build up in between the 206 support and skin layer interface. These aggregates then served as nucleation sites for the continued 207 deposition of other enzymes. As can be observed in Figure 1a&b, the skin layer is dense, and the pores 208 are too small for the ADH to pass through. In Phase 2, the first sublayer developed into multi-sublayers The permeate flux of PES and PVDF membranes decreases as the cumulative volume of permeate 232 increasing. This is because of the increasing of concentration polarization that happens on the surface 233 of the membrane. The concentration polarization is increase when the fouling on/in the membrane 234 increase (Giacobbo et al. 2018). In our study, this is desirable because it indicates that the enzyme is 235 successfully adsorbed and fouled the membrane. 236 237

Evaluation of membrane fouling 238
As described in the previous section, the governing mechanism of membrane fouling due to enzyme 239 immobilization procedure in this study is macromolecular adsorption. Two models were applied to 240 evaluate membrane fouling in this study, namely the flux recovery ratio (FRR) and Hermia's model. compared to PVDF with 66.1% (Figure 4). This could indicate that the fouling in PES is more critical 247 than PVDF, even though both membranes load almost the same amount of enzymes (Table 2). 248 As shown in Figure 4, PES and PVDF membranes have a higher percentage ratio of reversible resistance 249 with 39.24% and 12.25% respectively, compared to the irreversible resistance with 33.92% and 11.45% 250 respectively. Reversible resistance is where the fouling agents (ADH) that are loosely attached to the 251 membrane and possible to be desorbed to the bulk solution. On the other hand, irreversible fouling 252 indicates that the fouling agents are tightly bound to the membrane and no possibility of desorption to 253 the bulk solution. A lower total resistance ratio (Rt) will cause a higher FRR resulting in a lower total 254 flux loss as compared to the pristine membrane. 255 From the regression data summarized in Table 3, intermediate fouling (Figure 4b) is predicted using 256 Hermia's model for both membranes suggesting that each foulant has a probability to either deposit on 257 an unobstructed area of the membrane or deposit onto a previously deposited foulant particle (Kirschner 258 et al. 2019). This is in line with the estimation of total resistance where it was found that reversible 259 fouling is higher than irreversible fouling in PES and PVDF membrane.

Enzymatic membrane reactor performance 270
The permeate flux of PVDF membrane is higher compared to the permeate flux of PES membrane 271 during reaction ( Figure 5). Both permeate flux profiles are stable at and have the same value as permeate 272 flux at the end of Phase 3 during enzyme immobilization procedure (Figure 2). Even though the MWCO 273 of PES is larger than PVDF (Table 1) Cycle PES PVDF Figure 6 shows the percentage reaction conversion of formaldehyde to methanol for 5 cycles. The 281 average conversion for PES membrane is higher than PVDF membrane. ADH enzyme on the PES 282 membrane can retain its activity above 80% within five cycles while the ADH enzyme on the PVDF 283 membrane can retain its enzyme activity above 50%. In previous research, ADH retain its activity at 284 20.1% and 79.6% of its original activity on MNP-ADH and MGO-ADH membrane respectively (Liu 285 et al. 2015). Despite the amount of enzyme immobilized is almost the same for both membranes (Table  286 2), the productivity is not corresponded to it. Biocatalytic productivity for PES and PVDF membrane 287 is 5.2 and 4.7 mg methanol/mg ADH respectively. This is very much related to the arrangement of 288 enzymes in/on the membrane due to the enzyme-enzyme and enzyme-membrane interactions. Protein 289 molecules are surrounded by a hydration shell in solution (Chen and Sun 2003). In the presence of salt 290 (buffer) and ionizable chemical species, the hydration shell will be stripped off from the protein 291 molecule due to the hydration effect of the salt molecules of the protein environment (Lin et al. 2000). 292 This will result with an exposed area of hydrophobic zones of the enzymes making the hydrophobic 293 interactions between the ADH and the adsorbent surface (polypropylene) become stronger. The 294 hydrophobic interactions will induce conformational changes of the enzyme structure thus altering its 295 activity. It is also believed that high permeate flux during reaction by a more hydrophilic PVDF 296 membrane, thus some of the substrate not in contact with enzyme while passing through the membrane. 297 Slower permeate of PES during immobilization would distribute enzyme evenly on the polypropylene 298 fibre strands and skin-support interface, hence slower permeate flux during reaction ensuring optimum 299 contact time with substrate. 300 301

Conclusion 302
The fibrous structure of asymmetric membrane support layer has the potential to be exploited as the 303 enzyme immobilization matrix. Almost 100% of enzymes in the feed were successfully immobilized 304 following adsorption mechanism and charge interaction. A combination of concentration polarization 305 effect which is critical in dead end filtration method and convective transport, drive the mass transfer 306 of the enzymes and further, aided in the docking. There are three phases of permeate flux decline which 307 are strongly related to the process of enzyme adsorption mechanism in the membrane, following this 308 pattern: molecular sorption and particle depositiondevelopment of multi-sublayerssublayers 309 rearrangement and stabilization. Ionizable chemical components and membrane surface, van der Waals 310 force, isoelectric point are consolidated factors, responsible for the membrane charges interaction which 311 further aid in the enzyme's entrapment. Membrane fouling is described via Hermia's model and flux 312 recovery ratio. Both models conclude that intermediate fouling dominates. Reversible fouling is higher 313 than irreversible for both membranes, but the intensity of irreversible fouling is more prominent in PES, 314 indicating fouling is severe in PES. The system showed stable, high enzyme conversion of more than 315 80% of formaldehyde to methanol in five cycles. This promotes a good biocatalytic productivity data 316 for the enzymes proving that enzyme immobilization in the reverse asymmetric membrane feasible to 317 be applied in other biocatalytic reactions.