Copolymer membranes

Abstract

The invention relates to methods of making membranes by combining a polymer with one or more of a methacrylate monomer in conditions allowing copolymerisation to form a functional and selective membrane.

Description

[0001] COPOLYMER MEMBRANES

[0002] Field of the invention

[0003] The invention relates to copolymer membrane comprising methacrylates, and their use in downstream bioseparation of biomolecules.

[0004] Introduction

[0005] The high demand for large therapeutic biomolecules (e.g., monoclonal antibodies (mAbs)), adeno-associated virus (AAV) vectors), leads to a significant increase in upstream production, which challenges the downstream purification processes to keep up the pace. Currently, column chromatography is the leading technology for downstream protein purification, using resin beads with different functionalities adapted to the necessary process steps. However, the high mass transfer resistance and the high operating pressure add limitations to the separation capacity of column chromatography, resulting in low throughput and time-consuming operation. Moreover, the need for resin pre-packing and harsh cleaning conditions poses practical challenges and limits the process scalability.

[0006] As an alternative, membrane chromatography has been developed to operate at higher flow rates and to reduce the processing time. The lower pressure drop and high throughput potential of membrane-based operations are attributed to the convective mass transfer through the membrane matrix, instead of the significantly slower intra-particle diffusion of proteins within resin beads. In addition, membranes can be easily modularized into capsules or cartridges of various sizes offering flexibility for upscaling with a much smaller footprint, potentially reducing the operational cost. Amongst the membrane chromatography modes used in downstream protein purification, ion-exchange membrane chromatography (IEMC) has been utilized as a promising alternative in the process-scale separation of biomolecules in general, such as mAbs, especially in the polishing step, e.g., to remove impurity proteins from the mAb stream. The situation with viral vector purification is more urgent due to the much larger size of the viral particles that makes the use of resin impractical. Nevertheless, in general the commercial membrane used in IEMC has a lower binding capacity than column chromatography resins (<50 %) and is known to be less chemically stable, which remains the technical bottleneck for industrial-scale applications in biopharmaceutical production. To enhance the binding capacity, optimizing the ligand density and distribution during membrane fabrication is regarded as a useful strategy.

[0007] Besides the low binding capacity, the relatively wide pH operation window limits the application of IEMC. For example, to separate BSA from the protein mixture from biological fluids, the BSA proteins are adsorbed at pH 6.0 and eluted at pH 3.0 with the commercial Sartobind Q75 membrane adsorber. How to narrow the pH operation window of chromatographic membranes and achieve more precise capture-and- release of proteins remains a major interest, from the perspectives of both process efficiency (e.g., precision of capture-and-release, product quality) and environmental concerns (e.g., material consumption, equipment corrosion, waste discharge). Recently, pH-responsive polymers have been applied to improve the precision of antibody purification and were considered a useful strategy to narrow the pH operation window. It has been shown that the application of pH-responsive membranes can narrow the operation window (binding at pH 5.5, elution at pH 7), and the DBCio% of the membrane (210 mg / g) reached the same level as commercial resins (around 200 mg / g). However, the obtained optimal binding pH (or binding behavior) of the previous membrane of Gao et al. (e.g., pH 5.5) cannot be applied to capture proteins with different isoelectropoints. Also, the currently available commercial membranes were not suitable for the purification of viral vector capsids that have different physiochemical properties, especially concerning the minor charge difference between full and empty capsids. Therefore, customizing the binding behavior of membranes from chemistry design and fabricating high-performance (e.g., high binding capacity, antifouling) membranes by structure optimization are considered as two promising strategies for developing next-generation membrane chromatography for a wide range of applications.

[0008] PVDF-based copolymer membranes were disclosed in Gao et al. (2023) Chem. Engin.

[0009] J. 476, 146700.

[0010] SUMMARY OF THE INVENTION

[0011] Membrane chromatography is considered as a sustainable alternative for downstream purification to overcome the production bottleneck of biopharmaceuticals. However, two of the major factors impeding its industrial adoption are the lack of controllable functionalization and competitive binding capacity.

[0012] Herein, a series of pH-responsive copolymers was synthesized to manipulate the binding behavior of membranes, in which the composition of polymers was optimized by adjusting the type and dosage of functional monomers. With the application of coaxial electrospinning, membranes with core-shell fibre structure were fabricated, where the a functional copolymer was incorporated onto the fibre surface to reduce the material consumption while producing desirable membrane structure. The performance of these membranes will be highlighted by comparing their performance with reported resins and membranes.

[0013] A series of pH-responsive copolymers including poly(vinylidene fluoride)-graft- poly(2-diethylaminoethyl methacrylate) (PVDF-g-PDEA), poly(vinylidene fluoride)- graft-poly(2-dimethylaminoethyl methacrylate)-graft-poly(2-diethylaminoethyl methacrylate) (PVDF-g-PDMADEA), poly(vinylidene fluoride)-graft-poly(2- diisopropylaminoethyl methacrylate) (PVDF-g-PDPA), and poly(vinylidene fluoride)- graft-poly(2-dimethylaminoethyl methacrylate)-graft-poly(2-diisopropylaminoethyl methacrylate) (PVDF-g-PDMADPA) were synthesized to customize the pH-responsive behavior of membranes. Accordingly, four membranes with core-shell naofibre structure were developed via coaxial electrospinning. In each of these membranes, the polyacrylonitrile (PAN) polymer was used as a core material, and each copolymer was incorporated as the shell material, producing membranes with four different chemistries, namely membrane DEA@PAN, DMADEA@PAN, DPA@PAN, DMADPA@PAN, respectively. The static binding performance proved that the optimal pH (ranging from 4 to 6) for bovine serum albumin (BSA) capture can be controlled by manipulating the polymer chemistry. The static binding capacity of DEA@PAN, DMADEA@PAN, and DMADPA@PAN membranes reaches over 400 mg / g membrane, which is higher than that of the DPA@PAN membrane (around 350 mg / g). The dynamic binding capacity (DBCio% around 300 mg / g ) of DEA@PAN, DMADEA@PAN, and DMADPA@PAN membranes remained stable up to an order of magnitude flowrate than the conventional materials, e.g., 70 mL / min. Also the membrane performance can be recovered by cleaning with 0.2 M NaOH solution.

[0014] To evaluate the membrane performance in purifying other types of biomolecules, the membranes developed above, along with a DMA@PAN membrane prepared from our previously developed polymer (PVDF-g-PDMA), were applied for adeno-associated virus (AAV) purification. The developed membrane exhibits competitive performance in separating full and AAV capsids, a challenging step in AAV downstream processing due to the minimal differences in their surface charge and structure, despite their markedly different therapeutic efficacy and safety profiles. On the one hand, at pH 5.0, the DMA@PAN membrane (the PVDF-g-PDMA polymer was developed in Chem. Engin. J. 476, 146700) exhibited highly selective binding of empty capsids, with a significant decrease in their concentration after 24 hours (****p < 0.0001), while the full and partial capsid levels remained statistically unchanged. On the other hand, with the DEA@PAN and DMADEA@PAN membrane at pH 5.0, both empty and full capsid counts decreased significantly, respectively. This demonstrates that the DEA@PAN and DMADEA@PAN membrane binds not only empty capsids but also a large portion of full capsids under acidic conditions, which requires a subsequent step to isolate the full ones. However, these are also desired outcomes that can fill the current gaps of tailored membranes for viral vector separation.

[0015] The invention is further summarized in the following statements:

[0016] 1. A method of making a membrane comprising the steps of:

[0017] - a) combining a polymer with one or more of a methacrylate monomer in conditions allowing copolymerisation thereby forming a copolymer,

[0018] - b) applying the obtained copolymer as a shell material on nanofibres of a membrane.

[0019] 2. The method according to statement 1, wherein the methacrylate monomer is one or more of DMA (2-dimethylaminoethyl methacrylate), DEA (2-diethylaminoethyl methacrylate) and DPA (2-diisopropylaminoethyl methacrylate).

[0020] 3. The method according to statement 1 or 2, wherein the polymer is alkali treated polyacrylonitrile (PAN) or alkali treated Polyvinylidene difluoride (PVDF).

[0021] 4. The method according to any one of statements 1 to 3, wherein the copolymer is applied on an electrospun nanofibre.

[0022] 5. The method according to statement 4, wherein the core of the nanofibre is a polyacrylonitrile .

[0023] 6. The method according to any one of statements 1 to 5, wherein in step a) 1 mole of alkali treated PVDF is combined with between 0.05 and 0.5 mole of methacrylate.

[0024] 7. The method according to any one of statements 1 to 5, wherein in step a) 1 mole of alkali treated PVDF is combined with between 0.1 and 0.2 mole of methacrylate.

[0025] 8. The method according to any one of statements 1 to 6, wherein the alkali treated polymer is copolymerised with DEA.

[0026] 9. The method according to any one of statements 1 to 5, wherein the alkali treated polymer is copolymerised with DMA and DEA.

[0027] 10. The method according to statement 9, wherein in step a) the amount [mokmol] of DMA versus DEA is between 10: 1 and 1 : 10. 11. The method according to statement 9, wherein in step a) the amount [mokmol ] of DMA versus DEA is between 5: 1 and 1 :5.

[0028] 12. The method according to any one of statements 1 to 5, wherein the alkali treated polymer is copolymerised with DPA.

[0029] 13. The method according to any one of statements 1 to 5, wherein the alkali treated polymer is copolymerised with DMA and DPA.

[0030] 14. The method according to statement 13, wherein in step a) 1 the amount [mokmol] of DMA versus DPA is between 10: 1 and 1 : 10.

[0031] 15. The method according to statement 13, wherein in step a) 1 the amount [mokmol] of DMA versus DPA is between 5: 1 and 1 :5.

[0032] 16. A membrane made of nanofibrous core material and an outer shell wherein the outer shell is a copolymer grafted with one or more of a methacrylate.

[0033] 17. The membrane according to statement 16, wherein the methacrylate is one or more selected from the group consisting of DMA, DEA and DPA.

[0034] 18. The membrane according to statement 17, comprising between 1 methacrylate unit per 2 PVDF units to 1 methacrylate unit per 15 PVDF units.

[0035] 19. The membrane according to statement 17 or 18, comprising between 1 methacrylate unit per 5 PVDF units to 1 methacrylate unit per 10 PVDF units.

[0036] 20. The membrane according to any one of statements 17 to 19, wherein the outer shell of the fibre is a copolymer comprising DEA.

[0037] 21. The membrane according to any one of statements 17 to 19, wherein the outer shell is a copolymer DMA and DEA.

[0038] 22. The membrane according to statement 21, wherein the ratio between DMA and DEA units ) is between 10: 1 and 1 : 10.

[0039] 23. The membrane according to statement 21, wherein the ratio between DMA and DEA units) is between 5: 1 and 1 :5.

[0040] 24. The membrane according to any one of statements 17 to 19, wherein the outer shell is a copolymer comprising DPA. 25. The membrane according to any one of statements 17 to 19, wherein the outer shell is a copolymer comprising DMA and DPA.

[0041] 26. The membrane according to statement 25, wherein the ratio between DMA and DPA units is between 10: 1 and 1: 10.

[0042] 27. The membrane according to statement 25, wherein the ratio between DMA and DPA units is between 5:1 and 1 :5.

[0043] 28. The membrane according to any one of statements 15 to 27, which is copolymer of said methacrylate or said methacrylates with an alkali treated PVDF.

[0044] 29. Use of a membrane obtained by the method of any one of statements 1 to 14 or use of a membrane according to any one of statements 15 to 28 for the purification of biomolecules.

[0045] 30. The use according to statement 27, wherein the biomolecule is a protein or a polynucleotide.

[0046] 31. The use according to statement 27, wherein the protein is an antibody.

[0047] 32. A method of making a membrane comprising the steps of: a) combining a polymer with DEA or DPA in conditions allowing copolymerization, thereby obtaining a copolymer, and b) further processing the copolymer to form a membrane.

[0048] 33. The method according to statement 32, wherein the polymer is an alkali treated PVDF.

[0049] 34. The method according to statement 32, wherein the copolymer is purified from unreacted reagents and solvents, prior to step b).

[0050] 35. The method according to statement 32 or 33, wherein in step a) the alkali treated polymer is combined with a mixture of DEA and DMA.

[0051] 36. The method according to statement 32 or 33, wherein in step a) the alkali treated polymer is combined with a mixture of DPA and DMA. 37. The method according to any one of statements 32 to 36, wherein in step b) the membrane is formed using phase inversion.

[0052] 38. A method of making a membrane comprising the step of: combining two or more amino-functional methacrylate monomers in conditions allowing copolymerisation, thereby forming a homo- or copolymer of the methacrylates.

[0053] 39. The method according to claim 38, wherein the methacrylate monomers are selected from DMA (2-dimethylaminoethyl methacrylate), DEA (2-diethylaminoethyl methacrylate) and DPA (2-diisopropylaminoethyl methacrylate)

[0054] 40. The method according to claim 38 or 39, wherein the copolymerisation of the methacrylates occurs in combination with a further polymer.

[0055] 41. The method according to claim 40, wherein the further polymer is a partially defluorinated PVDF ( Polyvinylidene difluoride) comprising conjugated double bonds (C=C) along the backbone and optionally comprising hydroxyl (-OH) and carbonyl (C=O) groups.

[0056] 42. The method according to claim 41, wherein the PVDF is an alkali treated PVDF.

[0057] 43. The method according to claim 40, wherein the further polymer is polyacrylonitrile (PAN) comprising free radicals.

[0058] 44. The method according to claim 43, wherein the PAN is a PAN treated with irradiation or with an alkali.

[0059] 45 The method according to any one of claims 38 to 44, further applying the obtained copolymer as a shell material on nanofibres of a membrane.

[0060] 46. The method according to claim 45, wherein the copolymer is applied on an electrospun nanofibre,

[0061] 47. The method according to claim 45 or 46, wherein the core of the nanofibre is a polyacrylonitrile. 48. The method according to any one of claims 40 to 47, wherein 1 mole of the further polymer is combined with between 0.05 and 0.5 mole of methacrylate, or with between 0.1 and 0.2 mole of methacrylate.

[0062] 49. The method according to any one of claims 40 to 48, wherein the further polymer is copolymerised with DMA and DEA.

[0063] 50. The method according to claim 38 or 39, or any one of claims 45 to 49, wherein DMA and DEA are copolymerised, in the absence of a further polymer.

[0064] 51. The method according to claim 49 or 50, wherein the amount [mokmol] of DMA versus DEA is between 10: 1 and 1 : 10, or is between 5: 1 and 1 :5.

[0065] 52. The method according to any one of claims 40 to 48, wherein the further polymer is copolymerised with DMA and DPA.

[0066] 53. The method according to claim 38 or 39, or any one of claims 45 to 49, wherein DMA and DPA are copolymerised, in the absence of a further polymer.

[0067] 54. The method according to claim 52 or 53, wherein the amount [mokmol] of DMA versus DPA is between 10: 1 and 1 : 10, or is between 5: 1 and 1 :5.

[0068] 55. A membrane comprising a copolymer of two or more methacrylates, said copolymer optionally comprising a further polymer.

[0069] 56. The membrane according to claim 55, wherein the metacrylates are selected from the group consisting of DMA, DEA and DPA.

[0070] 57. The membrane according to claim 55 or 56, wherein the further polymer is PAN or PVDF.

[0071] 58. The membrane according to any one of claims 55 to 57, which is made of a nanofibrous core material and an outer shell containing the copolymer.

[0072] 59. The membrane according to any one of claims 57 or 58, comprising between 1 methacrylate unit per 2 vinylidene difluoride (VDF) units to 1 methacrylate unit per

[0073] 15 VDF units. 60. The membrane according to claim 58 or 59, wherein the outer shell is a copolymer of PVDF with DMA and DEA, or wherein the outer shell is a copolymer of PVDF with DMA and DPA.

[0074] 61. The membrane according to claim 58 or 60, wherein the outer shell is a copolymer of DMA and DEA in the absence of PVDF, or wherein the outer shell is a copolymer of DMA and DPA in the absence of PVDF.

[0075] 62. The membrane according to any one of claim 55 to 61, wherein the ratio between DMA and DEA units or the ratio between DMA and DPA units is between 10: 1 and 1 : 10 or is between 5: 1 and 1 :5.

[0076] 63. Use of a membrane obtained by the method of any one of claims 38 to 54 or of a membrane according to any one of the claim 55 to 62 for the purification of biomolecules, such as an antibody or a polynucleotide.

[0077] 64. Use of a membrane functionalised with one or more methacrylate monomers, for the purification of Viral Vector capsids.

[0078] 65. The use according to claim 64, wherein the one or more methacrylate monomers are selected from DMA (2-dimethylaminoethyl methacrylate), DEA (2- diethylaminoethyl methacrylate) and DPA (2-diisopropylaminoethyl methacrylate), .

[0079] 66. The use according to claim 27 or 28, wherein the membrane is functionalised with DMA (2-dimethylaminoethyl methacrylate), and the purification of the biomolecules is the separation of empty from full Adenovirus Associated viral vector capsids.

[0080] 67. A method of purifying Adenovirus Assisted Vector capsids comprising the steps of:

[0081] - applying a solution with a pH between 5 and 6, the solution comprising full and empty AAV capsids on a membrane functionalised with DMA (2-dimethylaminoethyl methacrylate),

[0082] - collecting the solution which flows through the membrane, the solution comprising the full AAV capsids, while the empty capsids are retained on the matrix. 68. A method of purifying Adenovirus Assisted Vector capsids comprising the steps of:

[0083] - applying a solution with a pH between 5 and 6, the solution comprising full and empty AAV capsids on a membrane functionalised with DEA and / or DPA, under conditions binding both full and empty AAV on the membrane,

[0084] - eluting full AAV capsids from the membrane by increasing the salt concentration and / or the pH of the solution.

[0085] Detailed description

[0086] Figure legends

[0087] Fig. 1 H1NMR spectra of the PVDF-g-PDEA, PVDF-g-PDMADEA, PVDF-g-PDPA, and PVDF-g-PDMADPA polymer.

[0088] Fig. 2 FTIR spectra of the DEA@PAN, DMADEA@PAN, DPA@PAN, and DMADPA@PAN membrane.

[0089] Fig. 3 The surface morphology of DEA@PAN, DMADEA@PAN, DPA@PAN, and DMADPA@PAN membranes.

[0090] Fig. 4 The TEM image of DEA@PAN, DMADEA@PAN, DPA@PAN, and DMADPA@PAN membranes.

[0091] Fig. 5 Membrane permeance and protein rejection of DEA@PAN, DMADEA@PAN, DPA@PAN, and DMADPA@PAN membranes.

[0092] Fig. 6. The effect of pH on static protein adsorption of DEA@PAN, DMADEA@PAN, DPA@PAN, and DMADPA@PAN membranes.

[0093] Fig. 7. The chromatogram of DEA@PAN, DMADEA@PAN, and DMADPA@PAN membranes under different flow rates (1 mg / mL BSA).

[0094] Fig. 8 The adsorption isotherms and kinetics of DEA@PAN, DMADEA@PAN, DPA@PAN, and DMADPA@PAN membranes.

[0095] Fig. 9. The protein recovery of DEA@PAN, DMADEA@PAN, and DMADPA@PAN membranes under different flow rates. Fig. 10. Breakthrough curve of DEA@PAN, DMADEA@PAN, and DMADPA@PAN membranes under different flow rates (BSA concentration = 5 mg / mL).

[0096] Fig. 11. Breakthrough curve of DEA@PAN, DMADEA@PAN, DPA@PAN, and DMADPA@PAN membranes under different flow rates (BSA concentration = 2 mg / mL).

[0097] Fig. 12. Antifouling performance test of DEA@PAN, DMADEA@PAN, DMADPA@PAN, and DPA@PAN membranes (BSA concentration = 1 mg / mL).

[0098] Fig. 13. Protein recovery in the antifouling performance test of DEA@PAN, DMADEA@PAN, and DMADPA@PAN membranes (BSA concentration = 1 mg / mL).

[0099] Fig. 14. Stability performance of DEA@PAN, DMADEA@PAN, DPA@PAN, and

[0100] DMADPA@PAN membranes in terms of chromatogram with 0.2 M NaOH cleaning.

[0101] Fig. 15. Stability performance of DEA@PAN, DMADEA@PAN, DPA@PAN, and

[0102] DMADPA@PAN membranes in terms of BSA recovery % with 0.2 M NaOH cleaning.

[0103] Fig. 16 Static binding of adeno-associated virus serotype 9 (AAV9) capsids on the DMA@PAN (DMA polymer developed in our previous research Chem. Engin. J. 476, 146700) membrane at pH 5.0 and pH 6.0. Capsid populations (empty, partial, full) were quantified at 0 h and after 24 h using mass photometry. Error bars represent standard deviations of triplicate measurements. Statistical analysis was performed using one-way ANOVA followed by Tukey s HSD test. Significance levels: *p < 0.05, **p < 0.01, ***p < 0.001,****p < 0.0001.

[0104] Fig. 17 Static binding of adeno-associated virus serotype 9 (AAV9) capsids on the DEA@PANmembrane at pH 5.0 and pH 6.0. Capsid populations (empty, partial, full) were quantified at 0 h and after 24 h using mass photometry. Error bars represent standard deviations of triplicate measurements. Statistical analysis was performed using one-way ANOVA followed by Tukey s HSD test. Significance levels: *p < 0.05, **p < 0.01, ***p < 0.001,****p < 0.0001.

[0105] Fig. 18 Static binding of AAV9 capsids on DMADEA@PAN membranes at pH 5.0 and pH 6.0. Capsid populations (empty, partial, and full) were quantified by mass photometry at 0 h and after 24 h of membrane incubation. Error bars represent standard deviations of triplicate measurements. Statistical analysis was performed using one-way ANOVA followed by Tukey s HSD test. Significance levels: *p < 0.05, **p < 0.01, ***p < 0.001,****p < 0.0001.

[0106] The methods of the invention allow to develop membranes or copolymers with tailored pH-responsive properties. The following parameters have an impact on pH response behavior of the membrane:

[0107] -The addition of a more hydrophobic ligand into the copolymer structure lowers the pKa (proton accepting capacity) of the copolymer / membrane, as compared to original polymer. When hybridizing with a more hydrophobic ligand, such as DEA or DPA with longer alky chain (hence more hydrophobic) being added to the copolymer with DMA in the structure, the copolymer and the resulting membrane would exhibit a lower apparent pKa. This is primarily due to the hydrophobic environment surrounding the amine groups, which impedes the diffusion of water and protons, thereby reducing the degree of protonation. Such "hydrophobic suppression" of protonation leads to a shift in the effective pKa to lower values compared to the free monomer in solution.

[0108] The incorporation of a more hydrophilic ligand into the copolymer structure will improve the charge density of the copolymer / membrane over a broader pH range. For example, hybridizing a more hydrophilic (e.g., DMA) and hydrophobic (e.g., DPA) ligands results in a synergistic enhancement of membrane performance. In detail, the more hydrophilic DMA promotes water uptake and increases proton accessibility, thereby improving the overall charge density of the membrane over a broader pH range (e.g., pH 4-6) compared to membranes functionalized with individual hydrophobic ligands (e.g., DEA@PAN and DPA@PAN). This hybrid ligand strategy facilitates more robust anion-exchange capabilities, especially for target proteins that carry negative surface charges only under mildly acidic conditions. By altering the ratio between different ligands and the grafting density will allow fine tuning of the pKa and surface charge density of the copolymer / membrane to meet specific separation requirements.

[0109] For example, to achieve sequential capture of different proteins and release without replacing the membrane, expanding the operational pH range through ligand hybridization (e.g., combining DMA with DEA) offers a promising solution.

[0110] The figures of this application illustrates various advantages of membranes of the present invention.

[0111] The H1NMR results in figure 1 prove the successful synthesis of the claimed copolymers.

[0112] The FTIR results of figure 2 prove the appearance of amine groups in the membrane, indicating that functional monomers has been immobilized on the membrane.

[0113] It is seen in figure 3 that the membranes exhibit cylindrical and straight fibres with a bead-free morphology, indicating that the introduction of functional polymers has little effect on the morphology of the electrospun fibres.

[0114] To further verify the 'core-shell'-like structure, TEM was applied to investigate the morphologies of membranes. Fig. 4 demonstrates that all the coaxial electrospun fibres exhibit an obvious 'core-shell'-like structure, which is distinguished via the difference in brightness level between the inner 'core' and the outer 'shell'. The PAN 'core' is homogeneously encapsulated by the PVDF-based functional polymer 'shell' along the axis.

[0115] All the membranes show in figure 5 a permeance higher than 13000 L m-2h1bar-1, and there is no significant difference between these four membranes. The rejection of BSA and IgG is lower than 5% with these membranes, avoiding the undesirable rejection of proteins during the applications.

[0116] As shown in Fig. 6, membranes showed a pH-responsive property for BSA adsorption. The binding capacity of IgG is insignificantly lower than that of BSA. The optimal pH for BSA binding decreased with the increase in the hydrophobicity of functional monomers grafted to the copolymer. Especially, the DMADEA@PAN membrane showed a relatively wide working window (pH 4-6) for maximum binding of BSA.

[0117] As shown in Fig. 7, increasing the flow rate from 1 to 70 mL / min did not have a significant effect on the chromatogram shape of DEA@PAN, DMADEA@PAN membranes, which means that the membrane performance is stable under the condition of high flow rate. However, the elution peak of the DMADPA@PAN membrane decreases with increasing flow rate, which might be caused by the more severe fouling under high flow rate.

[0118] The binding capacity of the membranes increases with increasing BSA concentration, and the binding capacity reaches a plateau when the BSA concentration is higher than 40 mg / L (see adsorption isotherm data in figure 8). With 100 mg / mL BSA, the binding capacity of the membranes in decreasing order follows: DMADPA@PAN > DMADEA@PAN> DEA@PAN> DPA@PAN. The adsorption kinetics indicated that the membranes with more hydrophobic monomers (e.g., DMADPA@PAN and DPA@PAN) shows a much slower adsorption rate.

[0119] Figure 9 shows that the increase of flow rate from ImL / min to 70 mL / min did not have a significant effect on the BSA recovery.

[0120] Because of the high binding capacity of membranes, the 10% breakthrough of the membranes cannot be observed with 1 mg / mL BSA solution (Figures 7 and 8). Therefore, the dynamic binding capacity of membranes with 2 mg / mL and 5 mg / mL BSA solution was evaluated respectively (Figures 10 and 11). With 2 mg / mL BSA solution, the dynamic binding capacity (DBCio%, flow rate = 20 mL / min) reaches 320 mg / g, 270 mg / g, and 295 mg / g with DMADEA@PAN, DEA@PAN, and DMADPA@PAN membranes, respectively. With 5 mg / mL BSA, the DBCio% (flow rate = 20 mL / min) of DEA@PAN, DMADEA@PAN and DMADPA@PAN membranes reaches 374 mg / g, 407 mg / g and 385 mg / g, respectively.

[0121] As shown in Figure 12, after 5 runs without intermediate cleaning, the shapes of the chromatograms from the DEA@PAN and DMADEA@PAN membranes do not change significantly, indicating good antifouling performance. In comparison, the elution peak of the DMADPA@PAN membrane drops significantly, which might be caused by the membrane fouling. The dynamic binding capacity of the DPA@PAN membrane is below 50 mg / g, which might be caused by its slow binding kinetic profile (Figure 8). Because the main advantage of membrane chromatography is its higher operation flow rate compared to resin columns, the further performance evaluation of the DPA@PAN membrane was not included.

[0122] Due to the antifouling property of DEA@PAN and DMADEA@PAN membranes, the protein recovery of these two membranes keeps stable (higher than 99%) after 5 runs (see Figure 13). However, the protein recovery of the DMADPA@PAN membrane decreases to 60% after 3 runs, possibly due to membrane fouling.

[0123] To evaluate the reusability of the membranes, membranes in chromatographic applications were tested to evaluate its stability via standard cleaning and regeneration procedure. The stability of the membranes through harsh regeneration conditions was not well explored in the state-of-the-art membranes for similar applications. Figures 14 and 15 shows the stability of the membrane in terms of the dynamic binding capacity at 10% breakthrough (i.e., DBCio%) and the recovery of BSA after at least 10 cycles of sample loading-washing-elution-cleaning, where cleaning utilized 0.2 mol / L NaOH solution. The chromatogram does not change significantly after 10 cycles. The NaOH cleaning also does not have a significant effect on the BSA recovery. These results indicate that the membrane developed in the present invention can be thoroughly cleaned with 0.2 mol / L NaOH, and the binding performance was not affected in a 10-cycle cleaning process.

[0124] The superior properties of membranes of the invention are compared with a representative commercial resin, a commercial membrane and with the membranes of Gao et al. cited above.

[0125] The dynamic binding performance of then newly developed membranes is compared with that of reported data including commercial resin and membranes, under various protein concentration from 1 to 5 mg / mL.

[0126] Table 1 Comparison of dynamic binding capacity with commercial resin and literature reported membranes for antibody purification

[0127] 1Resin commonly applied at flow rate < 2 mL / min;

[0128] 2Tested with lysozyme (Chiu et al. (2011) Polymer Letters 5, 308-317). Commercial membrane tested up to range of 24-48 mL / min;

[0129] 3Membranes made by Gao et al. (2023), Chem. Engin. J. 476, 146700) tested in the range of 1-10 mL / min;

[0130] 4~5Membranes in this invention all stably evaluated in the range of 20-70 mL / min.

[0131] * mg / mL referred to mg biomolecule per mL adsorbent. mg / g referred to mg biomolecule per g adsorbent Figure 16 shows that the DMA@PAN membrane exhibited highly selective binding of empty capsids at pH 5, whereas a loss of selectivity can be observed at pH 6 because both empty and full capsid can be captured. This selectivity is consistent with the higher isoelectric point (pl) of empty capsids compared to full capsids. At pH 5.0, both capsid types are below their pl and thus positively charged, but full capsids (with lower pl) carry a stronger positive charge, which leads to greater electrostatic repulsion from the positively charged membrane surface. Empty capsids, being closer to their pl, are less positively charged at pH 5.0, allowing for reduced repulsion and stronger interaction, resulting in their preferential binding. At pH 6.0, both capsid types experience less electrostatic repulsion and can bind to the positively charged membrane. As a result, the membrane captures both empty and full capsids, reducing its ability to selectively remove the empty species.

[0132] As shown in Figure 17, all capsid types exhibited significant reductions after 24 hours of incubation at pH 5 and pH 6, indicating strong but non-selective binding behavior of the DEA@PAN membrane. At pH 5.0, empty capsid counts decreased significantly from 656 to as low as 50 particles, while partial and full capsid counts also dropped from 117 to around 34 and from 1033 to around 180, respectively. This demonstrates that the DEA@PAN membrane binds not only empty capsids but also a large portion of full capsids under acidic conditions. A similar pattern was observed at pH 6.0, with statistically significant reductions for empty, partial, and full capsids. These results confirm that selectivity is not pH-dependent in the case of the DEA@PAN membrane. Instead, the copolymer composition appears to promote broad-spectrum retention of AAV capsids.

[0133] As shown in Figure 18, all capsid types exhibited significant reductions after 24 hours of incubation at pH 5 and pH 6, indicating strong but non-selective binding behavior of the DMADEA@PAN membrane. At pH 5.0, empty capsid counts decreased significantly from 402 to as low as 80 particles, while partial and full capsid counts also dropped from 279 to around 64 and from 1055 to around 119, respectively. This demonstrates that the DMADEA@PAN membrane binds not only empty capsids but also a large portion of full capsids under acidic conditions. A similar pattern was observed at pH 6.0, with statistically significant reductions for empty, partial, and full capsids. These results confirm that selectivity is not pH-dependent in the case of the DMADEA@PAN membrane. Instead, the copolymer composition appears to promote broad-spectrum retention of AAV capsids.

[0134] Table 2 summarizes the performance of the DMA@PAN and DMADEA@PAN membranes in comparison to commercially available AAV purification technologies. The DMA@PANmembrane demonstrates strong selective binding, achieving approximately 95% full capsid recovery through preferential empty capsid depletion at mildly acidic pH (5.0-6.0), a range not typically addressed by commercial products. While the DMADEA@PAN and DEA@PAN membranes also provide high binding capacity, its full / empty discrimination was not explicitly optimized, making it potentially suitable for high-capacity early capture steps. In contrast, commercial affinity resins such as POROS CaptureSelect AAVX offer high binding capacities (>1013vp / mL) but do not explicitly address full / empty separation in published performance data. Ion exchange-based systems like Capto™ Q and Mustang® Q membranes achieved moderate enrichment of full capsids but require higher pH ranges (>8.0), which may pose stability challenges for certain serotypes. CIMmultus® QA monoliths offer improved separation (up to 80% full capsid recovery) but operate closer to the isoelectric points of the capsids, requiring tighter process control that might not be practical. Taken together, these results highlight that the DMA@PAN membrane uniquely fills a gap in the current purification landscape, providing a one-step selective empty capsid removal under acidic conditions, which can be particularly valuable to significantly improve the efficiency of in downstream purification workflows.

[0135] Traditional chromatographic purification of AAV vectors, including AAV9, typically involves multistep protocols comprising sequential binding, washing, and gradient or stepwise elution phases. These workflows often require careful optimization of salt concentrations, buffer compositions, and pH conditions to achieve differential elution of full and empty capsids, which possess nearly identical surface properties. Such strategies rely heavily on fine-tuned elution profiles— commonly using high salt or expensive ligands— to displace bound capsids from the stationary phase, increasing process complexity and cost. In contrast, the functionalized membranes described in this invention, particularly those based on DMA chemistry, enable a simplified one- step strategy. Under appropriately selected pH conditions, the DMA membranes preferentially capture empty capsids while allowing full capsids to flow through, eliminating the need for additional elution or regeneration steps. This significantly reduces processing time and buffer usage, while enhancing scalability and robustness of the purification process.

[0136] Table 2. Comparative Static Binding Performance of AAV Purification Technologies [1] Cell Gene Ther. Insights 4(7) (2018) : 637-645. doi : 10. 18609 / cgti.2018.061

[0137] [2] https: / / www.cytivalifesciences.com / en / us / knowledge-center / enhanced-aav- downstream- processing?srsltid=AfmBOoqE4juE9bbe_nBJxro6mHaupnQBOAEszoYnMcYMvYPJ9a7gkwNR

[0138] [3] Biotechnology Journal. 2022; 17(e2100219). doi : 10. 1002 / biot.202100219 [4] Journal of Chromatography A, 1740, 465586. doi : 10. 1016 / J. CHROMA.2024.465586 To benchmark the performance of our functionalized membranes, we compared their dynamic binding capacities (DBC) and elution strategies with established ionexchange (IEX) membranes and resins reported in literature (Table 3). Our DMA@PAN, DEA@PAN, and DMADEA@PAN membranes all exhibited high binding capacities (~2-3 x 1013vp / mL) within a mild pH range (5.0-6.0), with the DMA membrane demonstrating ~95% full capsid recovery without the need for an elution step. In contrast, commercial IEX platforms such as the AEX-TEA nonwoven membrane showed a binding capacity of 9.6 x 1013vp / mL at pH 8.0, requiring a multi-step NaCI elution. Similarly, Nuvia aPrime 4A and Nuvia HP-Q resins achieved binding capacities of 3 x 1013and 2.4 x 1013vp / mL, respectively, under alkaline conditions (pH 9.0) and employed multi-step salt elution protocols. These comparisons highlight the competitive binding performance of our membranes under milder conditions with simplified process requirements. Also, the DBC was not yet optimized, as it also depends on the viral particle concentration of the culture samples. However, initial tests using a defined AAV batch showed competitive binding performance to commercial products.

[0139] Table 3. Comparative Dynamic Binding Performance of AAV Purification Technologies

[0140] [1] Biotechnology Journal, 17(12), 2200211. https: / / doi.org / 10.1002 / biot.202200211 [2] Journal of Biological Engineering, 18(1), 6. https: / / doi.org / 10.1186 / sl3036-

[0141] 024-00409-4 Table 4 Comparison of dynamic binding capacity with commercial resin and literature reported membranes for antibody purification.

[0142] 1Resin commonly applied at flow rate < 2 mL / min;

[0143] 2Tested with lysozyme (Chiu et al. (2011) Polymer Letters 5, 308-317). Commercial membrane tested up to range of 24-48 mL / min;

[0144] 3Membranes made by Gao et al. (2023), Chem. Engin. J. 476, 146700) tested in the range of 1-10 mL / min;

[0145] 4~5Membranes in this invention all stably evaluated in the range of 20-70 mL / min.

[0146] * mg / mL referred to mg biomolecule per mL adsorbent.

[0147] ** mg / g referred to mg biomolecule per g adsorbent

[0148] The invention discloses the membrane preparation and evaluation of a series of pH- responsive membrane chromatography. To develop membranes with customized binding behavior, the pH-responsive property of membranes was manipulated by adjusting the composition of functional copolymers. With the application of coaxial electrospinning, nanofibrous membranes with a high specific area were fabricated, which enhances the interaction between the target protein and ligand. The resultant membranes showed a high water permeance (>13000 L m-2h1bar1) and low rejection rate (<5%) towards the model proteins (e.g., BSA and IgG). The static binding capacity of DEA@PAN, DMADEA@PAN, and DMADPA@PAN membranes reached over 400 mg / g, which is higher than that of the DPA@PAN membrane (around 350 mg / g). The dynamic binding capacity of DEA@PAN, DMADEA@PAN, and DMADPA@PAN membranes (DBCio% around 300 mg / g) remained stable in a wide range of operating flowrate. The ability of this newly developed membrane for BSA separation from a mixture of BSA and IgG was proved with binding-elution chromatography operation and SDS-PAGE analysis. The DMA@PAN membrane offers high selectivity for empty AAV capsids, making it particularly advantageous for applications where precise removal of non-functional empty particles is required to enhance therapeutic vector purity. In contrast, the DMADEA@PAN and DEA@PAN membrane, with its higher overall binding capacity and broader retention profile of all capsids, may be better suited for early capture steps, where maximizing viral particle recovery is prioritized over capsid differentiation. With its relatively stable physicochemical and mechanical properties, the membrane also presented robust performance in consecutive multicycle bind - elute - regeneration operations. The promising biocompatibility of the developed membrane indicated its potential to be used in biopharmaceutical / biomedical applications. The results obtained herein provide a new route for tailoring stimuli-responsive membranes and engineering highly porous pore structure of membrane adsorbers for highly efficient downstream purification of biomolecules in general. The chemistries provide a versatile membrane platform that can be tailored to fit different stages of AAV purification workflows, depending on whether the goal is selective enrichment or high-yield recovery.

[0149] Example 1 Membrane synthesis

[0150] The PVDF polymer (Solef 1015, MW 570-600 kDa) is from Solvay specialty polymers in Italy. KOH in solid form and absolute ethanol from Merck were used as reagents for the alkali treatment of PVDF. 2-(dimethylamino)ethyl methacrylate (DMA), 2- (Ddiethylamino)ethyl methacrylate (DEA), 2-(diisopropylamino)ethyl methacrylate (DPA), 2,2'-azobis(2-methylpropionitrile) (AIBN), dimethylformamide (DMF), polyacrylonitrile (PAN), and dimethylacetamide (DMAc) were purchased from Sigma Aldrich for copolymer synthesis and membrane fabrication. BSA from Sigma Aldrich and the NZ Bovine IgG from MP Biomedicals were used as model impurity and model mAbs, respectively. tris(hydroxymethyl)aminomethane hydrochloride (Tris-HCI), NaCI, NaOH, and HCI from Sigma Aldrich were used for the preparation of buffer solutions for binding tests. The initiator AIBN was purified by recrystallization from methanol. DMA, DEA, and DPA monomers were purified to remove inhibitors by filtration through a column made of basic alumina oxide powder. All other chemical agents were used as received without further purification.

[0151] The pH-responsive copolymer with different responsive behaviors was synthesized with free radical polymerization. To obtain the alkali-treated PVDF, PVDF powder was added to a beaker containing a mixture of 2.5 M KOH solution and ethanol at a ratio of around 35: 1. The solution was then heated at 60 °C for 20 min while stirring. The precipitated alkali-treated product was collected by filtration and then rinsed at least three times with deionized (DI) water, the collected product was dried in a vacuum oven at 60 °C for 24 h. Before adding monomers and initiator, a certain amount of alkali-treated PVDF powder was fully dissolved in DMF at 60 °C in a round-bottom flask under stirring. The mixture was cooled to room temperature. Then, a predetermined amount of monomer and initiator AIBN was added to the reaction solution under N2 atmosphere under continuous stirring. The polymerization was performed in an oil bath at 70 °C for 8 h. At the end of the reaction, the copolymer was precipitated in pure ethanol, filtered, and washed thoroughly with DI water. The resulting copolymer was dried completely in an oven at 60 °C for 24 h. The molar ratio between different monomers was adjusted to obtain a series of copolymers, which are named PVDF-g-DEA (100% DEA), PVDF-g-DPA (100% DPA), PVDF-g- DMADEA (DMA:DEA=1 :1), PVDF-g-DMADPA (DMA:DPA=1 :1). The ratio here is on a molar basis. Coaxial electrospinning was achieved with a co-axial needle made of two concentric stainless-steel needles, one of which became the outer needle delivering the dope for the shell material of the fibre; while the other is the inner needle that delivered the dope for the core material of the fibre. To obtain the dope sheath solution for the shell side, a copolymer with a concentration of 10 wt% was dissolved in DMAc. The pumping rate of the core and shell dopes was set to 1.5 mL / h and 1.0 mL / h, respectively. The applied voltage was 16 kV and the distance from tip to collector was about 25 cm. The humidity and ambient temperature were controlled at 45% and room temperature. Finally, the prepared membranes were dried at room temperature for 24 h to remove the remaining solvent. Membranes fabricated with different copolymers were named as DEA@PAN, DPA@PAN, DMADEA@PAN, DMADPA@PAN, respectively.

[0152] Example 2. Membrane characterization

[0153] The synthesized copolymers were characterized via proton nuclear magnetic resonance (XH NMR) spectroscopy that was performed on a 400 MHz Bruker instrument (Ettlingen, Germany) using acetone-d6 as the solvent. Attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR, PerkinElmer Spectrum 100, Germany) was used to analyze the chemical structure of copolymers. The surface and cross-sectional morphologies of membranes were characterized by scanning electron microscopy (SEM) (JEOL JSM-6010LV, Tokyo, Japan). Fourier transform infrared spectroscopy - attenuated total reflectance (FTIR-ATR) and a Kratos Axis Supra X-ray Photoelectron Spectrometer (XPS) were used for analyzing the chemical compositions of the membrane surface. The zeta potential of the membranes was analyzed using an Electrokinetic Analyzer (SurPASS 3, Anton Paar, Austria). The Brunauer-Emmett-Teller (BET) surface area of the membranes was measured at 77 K using a nitrogen adsorption-desorption Micromeritics ASAP-2020 analyzer (Micromeritics Co., USA). Example 3 Membrane performance testing

[0154] Solute rejection with single component solutions

[0155] The rejection of model biomolecules (i.e., IgG and BSA) with the developed membranes was evaluated in a lab-made cross flow filtration system with an effective membrane area of 7 cm2. Two simulated protein solutions of IgG (1 mg / mL) and BSA (1 mg / mL), respectively, were used for the test at room temperature under 1 bar. The permeance was determined by the volume of permeate solution collected at a given time interval, divided by the effective area of the tested membrane and the applied transmembrane pressure, and was expressed as L m-2h-1bar-1. The protein rejection (%) was calculated using the following equation:

[0156] R = (l-Cp / Cf)xlOO% (1) where the CPand C? are solute concentrations in the permeate and the feed solutions, in mg / mL, respectively. The protein concentrations for BSA and IgG were determined using the protocol of the bicinchoninic acid (BCA) protein assay kit. All experiments were repeated three times to ensure reproducibility.

[0157] Static binding performance

[0158] The static binding capacity of membranes was evaluated with two model proteins BSA and IgG, respectively, at six different pH values in the range of pH 3.0-8.0. Briefly, the respective stock solutions of BSA and IgG at a protein concentration of 2 mg / mL in Tris-HCI (10 mM) buffer were prepared, at varying pH between 3.0 and 8.0 with a pH interval of 1.0. Before static binding, the developed membranes were cut into coupons with a fixed weight (e.g., 12 cm2) and placed in 20 mL of Tris-HCI (10 mM) stock buffers for equilibration overnight. The membrane was then rinsed with Tris-HCI (10 mM) buffer and incubated in the protein stock solutions for 24 h in a shaker (25 °C, 100 rpm). After that, the membrane was taken out, and the protein concentration of the solution before and after binding was determined using the BCA assay. The binding capacity (Qe, in mg protein per g membrane, hereafter simplified as mg / g) was calculated by the following equation: where Co (mg / mL) and Ce(mg / mL) are the concentrations of initial protein solution and protein solution after adsorption, respectively, V (mL) is the protein solution volume, and m (g) is the membrane weight. All tests were repeated 3 times to ensure reproducibility.

[0159] Protocol of Dynamic binding capacity (DBC) measurements

[0160] The dynamic binding performance was evaluated with the Bio-Rad NGC100 Chromatography System. Membranes were housed into a membrane module printed in-house by a 3-D printer, and the detail of the membrane module can be found in Goa et al (cited above). Three buffers were used for the DBC tests, including the equilibrium buffer (namely buffer A), elution buffer (namely buffer B), and binding buffer. The buffer A is 10 mM Tris-HCI solution. The buffer B is 1 mol / L NaCI in Tris- HCI solution (10 mM, pH = 7.0). The binding buffer was prepared by dissolving a predetermined amounts of proteins (BSA and IgG) in 10 mM Tris-HCI solution. The respective pH of buffer A, buffer B and binding buffer was determined based on result of static binding test, and which was adjusted with the HCI and NaOH solutions where necessary.

[0161] Each DBC experiment involved an automatic four-step program: (1) In the equilibration step, 10 mL of buffer A was used to equilibrate the membranes at 1 mL / min. (2) In the sample application step, protein solution (e.g., 1 mg / mL BSA in buffer A) was injected. (3) In the membrane washing step, buffer A was used to remove the unbound protein. (4) In the elution step, buffer B was used to elute the bound protein. During the entire run, the protein concentration was monitored in situ by a UV spectrophotometer at a wavelength of 280 nm. The sample fractions (solution pass through the membranes) were collected and analyzed with a BCA protein assay kit. The DBC of the chromatographic membrane is the amount of the total protein absorbed into the membrane, before the occurrence of a significant breakthrough in the curve. The DBCio% was determined by measuring the dynamic binding capacity of the membranes at 10 % breakthrough. The equations for calculating the protein recovery and DBCio% are shown below. To evaluate the effect of flow rate and protein concentration on the dynamic binding capacity of the membranes, a breakthrough analysis was conducted at different flow rates and protein concentration ranging from 1 to 70 mL / min and 1 to 5 mg / mL, respectively.

[0162] The recovery of protein: where Co, Cbt, and Cerepresent the initial feed BSA concentration (mg / mL), the breakthrough BSA concentration (mg / mL), and the eluted BSA concentration (mg / mL), respectively. Vo, Vbt, and Veare the injected BSA volume (mL), volume where BSA breakthrough occurs (mL), and eluted volume (mL), respectively. Breakthrough concentration and volume are the concentration and volume of the fractions collected during the sample application step.

[0163] The dynamic binding capacity at 10% (DBCio%, mg / g): where Vbreak is protein solution volume (mL) when outlet concentration reached 10% of the original protein concentration. The Co is the feed protein concentration (mg / mL), and Vdead and m are the system dead volume (mL) and the membrane weight (g), respectively. Multi-cycle consecutive operation without cleaning

[0164] To evaluate the change of membrane performance without intermediate cleaning, the membrane was tested over five consecutive operation cycles. Each cycle followed the same protocol as demonstrated in the Section of "Protocol of Dynamic binding capacity (DBC) measurements". The operation of a subsequent run started after the elution step of the previous run, without any intermediate cleaning of the membrane.

[0165] Membrane regeneration

[0166] To evaluate the stability of membrane performance in industrial regeneration protocol, a 10-cycle experiment with alkaline cleaning was conducted. Each cycle was carried out as follows: Firstly, a dynamic protein binding run was conducted as described in the Section of "Protocol of Dynamic binding capacity (DBC) measurements". Secondly, 0.2 M NaOH solution was used for the membrane regeneration for 10 min at a flow rate of 1 mL / min. Thirdly, the NaOH solution in the system was removed with buffer A. The recovery and DBCio% were measured after each cycle and compared to quantify the changes in performance after repeated cycles.

[0167] Adsorption isotherm measurements

[0168] The membrane with a fixed weight was incubated in 10 mL of the BSA solution for 24 h in a shaker bath (25 °C, 100 rpm), the BSA concentration was set to 5, 10, 20, 40, 80 100 mg / L. The binding capacity of membranes in different concentrations was calculated following Eq. 2. The data was fitted to the Langmuir and Freundlich isotherms shown below: Langmuir isotherm:

[0169] Ce / Qe=1 / KdQm + Ce / Qm ( 5 ) where Ce(mg / mL) is the equilibrium protein concentration, Qe(mg / g) is the protein uptake at Ce, Qm is the protein saturated uptake (mg / g) and Kd is the dissociation constant (mL / mg).

[0170] Freundlich isotherm:

[0171] Qe =FCe1 / n where Qeand Cehave the same definitions as in Eq. 5; KF ((mg g-1) (L mg-1)1 / n) and n are the Freundlich constants. All tests were repeated 3 times to ensure reproducibility.

[0172] Adsorption kinetics study

[0173] To understand the adsorption kinetics of the membrane, membrane coupons with a fixed area were placed into a series of glass vials. Each vial contained one membrane coupon and 10 mL of the BSA solution with a pre-determined concentration. The vials were then incubated in a shaker bath (25 °C, 100 rpm) for 24 h. Regular sampling of the solution from the glass vial at a predetermined time points (e.g., 0.5, 1, 2, 3, 4, 5, 6, 7, 24 h) was performed to measure the change in the protein concentration of the solution due to adsorption. All liquid samples were measured to count the protein amounts through the BCA assay kit method. The experimental data of adsorption kinetics were fitted by the pseudo-first-order model and pseudo-second- order model, respectively. The equations of these models are shown below.

[0174] The pseudo-first-order model:

[0175] The pseudo-second-order model: <8)where Qe and Qt are the adsorption capacity at equilibrium and at any time t, respectively (mg g-1). ki (h-1) and k? (g mg1min-1) are the rate constants of the kinetic models. All kinetics experiments were repeated 3 times to ensure reproducibility. Membrane performance for AAV9 binding

[0176] To assess the AAV9 binding capacity of functionalized membranes, a static binding protocol was employed. Electrospun membranes including DMA@PAN, DEA@PAN and DMADEA@PAN were individually incubated in AAV9-containing solutions at two different pH conditions: pH 5.0 and pH 6.0. These pH levels were chosen to modulate electrostatic interactions between the membrane surface and AAV capsids, based on the known isoelectric point differences between full and empty capsids. The membranes were submerged in AAV9 solutions under static conditions, and samples of the supernatant were collected at two time points: immediately before membrane immersion (0 h) and after 24 hours (24 h). The amount of AAV9 remaining in solution was quantified using mass photometry (MP), a label-free single-molecule technique that measures the mass of individual viral particles in solution by detecting the light scattering contrast generated upon landing on a glass surface.

[0177] All measurements were performed in triplicate to ensure reproducibility. Statistical analysis was conducted using one-way analysis of variance (ANOVA) followed by Tukey s Honest Significant Difference (HSD) post hoc test to evaluate the significance of differences between groups. The statistical analysis was based on individual counts of empty, partial, and full capsids in the samples, as determined by MP, enabling quantitative comparison of membrane performance in selectively retaining viral particles. This allowed us to determine the impact of membrane chemistry and pH on AAV binding performance with statistical rigor.

[0178] To assess the AAV9 binding capacity of functionalized membranes, a dynamic binding protocol was employed. Membranes including DMA@PAN, DEA@PAN and DMADEA@PAN were placed in a membrane module and connected to the Bio-rad NGC chromatography system. Equilibration of the membrane was done with 10 mM Tris buffer at the desired pH (5.0 or 6.0). AAV-containing samples were loaded at a constant flow rate 1.5ml / min. Breakthrough curves were monitored in real time at 280 nm and 260 nm to track empty and full particles, respectively. The DBCio% was determined by measuring the dynamic binding capacity of the membranes at 10 % breakthrough.

Claims

CLAIMS1. A method of making a membrane comprising the step of: combining two or more amino-functional methacrylate monomers in conditions allowing copolymerisation, thereby forming a homo- or copolymer of the methacrylates.

2. The method according to claim 1, wherein the methacrylate monomers are selected from DMA (2-dimethylaminoethyl methacrylate), DEA (2- diethylaminoethyl methacrylate) and DPA (2-diisopropylaminoethyl methacrylate)3. The method according to claim 1 or 2, wherein the copolymerisation of the methacrylates occurs in combination with a further polymer.

4. The method according to claim 3, wherein the further polymer is a partially defluorinated PVDF ( Polyvinylidene difluoride) comprising conjugated double bonds (C=C) along the backbone and optionally comprising hydroxyl (-OH) and carbonyl (C=O) groups.

5. The method according to claim 4, wherein the PVDF is an alkali treated PVDF.

6. The method according to claim 3, wherein the further polymer is polyacrylonitrile (PAN) comprising free radicals.

7. The method according to claim 6, wherein the PAN is a PAN treated with irradiation or with an alkali.

8. The method according to any one of claims 1 to 7, further applying the obtained copolymer as a shell material on nanofibres of a membrane.

9. The method according to claim 8, wherein the copolymer is applied on an electrospun nanofibre,10. The method according to claim 8 or 9, wherein the core of the nanofibre is a polyacrylonitrile.

11. The method according to any one of claims 3 to 10, wherein 1 mole of the further polymer is combined with between 0.05 and 0.5 mole of methacrylate, or with between 0.1 and 0.2 mole of methacrylate.

12. The method according to any one of claims 3 to 11, wherein the further polymer is copolymerised with DMA and DEA.

13. The method according to claim 1 or 2, or any one of claims 8 to 12, wherein DMA and DEA are copolymerised, in the absence of a further polymer.

14. The method according to claim 12 or 13, wherein the amount [mokmol] of DMA versus DEA is between 10: 1 and 1 :10, or is between 5: 1 and 1 :5.

15. The method according to any one of claims 3 to 11, wherein the further polymer is copolymerised with DMA and DPA.

16. The method according to claim 1 or 2, or any one of claims 8 to 12, whereinDMA and DPA are copolymerised, in the absence of a further polymer.

17. The method according to claim 15 or 16, wherein the amount [mokmol] ofDMA versus DPA is between 10: 1 and 1 : 10, or is between 5: 1 and 1 :5.

18. A membrane comprising a copolymer of two or more methacrylates, said copolymer optionally comprising a further polymer.

19. The membrane according to claim 18, wherein the metacrylates are selected from the group consisting of DMA, DEA and DPA.

20. The membrane according to claim 18 or 19, wherein the further polymer is PAN or PVDF.

21. The membrane according to any one of claims 18 to 20, which is made of a nanofibrous core material and an outer shell containing the copolymer.

22. The membrane according to any one of claims 20 or 21, comprising between 1 methacrylate unit per 2 vinylidene difluoride (VDF) units to 1 methacrylate unit per 15 VDF units.

23. The membrane according to claim 21 or 22, wherein the outer shell is a copolymer of PVDF with DMA and DEA, or wherein the outer shell is a copolymer of PVDF with DMA and DPA.

24. The membrane according to claim 21 or 23, wherein the outer shell is a copolymer of DMA and DEA in the absence of PVDF, or wherein the outer shell is a copolymer of DMA and DPA in the absence of PVDF.

25. The membrane according to any one of claim 18 to 24, wherein the ratio between DMA and DEA units or the ratio between DMA and DPA units is between 10: 1 and 1 :10 or is between 5: 1 and 1 :5.

26. Use of a membrane obtained by the method of any one of claims 1 to 17 or of a membrane according to any one of the claim 18 to 25 for the purification of biomolecules, such as an antibody or a polynucleotide.

27. Use of a membrane functionalised with one or more methacrylate monomers, for the purification of Viral Vector capsids.

28. The use according to claim 27, wherein the one or more methacrylate monomers are selected from DMA (2-dimethylaminoethyl methacrylate), DEA (2-diethylaminoethyl methacrylate) and DPA (2-diisopropylaminoethyl methacrylate), .

29. The use according to claim 27 or 28, wherein the membrane is functionalised with DMA (2-dimethylaminoethyl methacrylate), and the purification of the biomolecules is the separation of empty from full Adenovirus Associated viral vector capsids.

30. A method of purifying Adenovirus Assisted Vector capsids comprising the steps of:- applying a solution with a pH between 5 and 6, the solution comprising full and empty AAV capsids on a membrane functionalised with DMA (2- dimethylaminoethyl methacrylate),- collecting the solution which flows through the membrane, the solution comprising the full AAV capsids, while the empty capsids are retained on the matrix.

31. A method of purifying Adenovirus Assisted Vector capsids comprising the steps of: - applying a solution with a pH between 5 and 6, the solution comprising full and empty AAV capsids on a membrane functionalised with DEA and / or DPA, under conditions binding both full and empty AAV on the membrane,- eluting full AAV capsids from the membrane by increasing the salt concentration and / or the pH of the solution.

Images