Purification of viruses, virus-like particles, and spherical biomolecules by filtration-assisted polyalkylene glycol precipitation

Abstract

The present invention relates to a method for purifying viruses, virus-like particles, extracellular vesicles, proteins having a diameter of 5 nm or greater, protein complexes having a diameter of 5 nm or greater, nucleic acids having a diameter of 5 nm or greater, protein-nucleic acid complexes having a diameter of 5 nm or greater, and / or molecules having a diameter of 5 nm or greater by filtration-assisted polyalkylene glycol precipitation, and a filtration module that can be used in the method.

Description

[Technical Field]

[0001] The present invention relates to a method for purifying viruses, virus-like particles, extracellular vesicles, proteins having a diameter of 5 nm or greater, protein complexes having a diameter of 5 nm or greater, nucleic acids having a diameter of 5 nm or greater, protein-nucleic acid complexes having a diameter of 5 nm or greater, and / or molecules having a diameter of 5 nm or greater by filtration-assisted polyalkylene glycol precipitation, and a filtration module that can be used in the method. [Background technology]

[0002] Viral vectors are increasingly playing a role as vectors in gene therapy and / or cell therapy. For example, lentiviral vectors (LVs) are one of the three viral vector systems most widely used in clinical trials for gene therapy and gene-modified cell therapy. Their relatively low stability poses challenges for downstream processing. Therefore, there is a constant need for the development and improvement of gentle process steps. While steric exclusion chromatography (SXC) has attracted interest in the field of viral vector purification, it has not yet been scaled up, raising the question of how this can be achieved. Another major viral vector used in gene therapy is the adeno-associated vector (AAV).

[0003] LVs have been used for many years in the biopharmaceutical industry, primarily in gene-modified cell therapy. Stable integration of the LV genome and long-term transgene expression have led to favorable therapeutic outcomes for certain diseases, such as acute lymphoblastic leukemia (ALL). The first pediatric ALL patient treated with LV-based gene-modified cell therapy is now cancer-free for 10 years. In clinical trials, LVs have been used to treat a wide range of diseases, including cancer, immune disorders, metabolic disorders, and rare congenital diseases. New potential applications for LVs are emerging. Recently, the use of LVs has gained importance not only as a potential vaccination platform using integrated LVs targeting infectious diseases, but also as a potential vaccination platform using non-integrated LVs. The wide range of diseases that can be treated with LVs and the emergence of new applications have led to an increased need for efficient LV bioprocessing. Many challenges remain in the production of LVs, and further optimization of purification, in particular, is needed.

[0004] The use of SXC to purify LV has recently been reported. To date, various viruses, including baculovirus, orf virus, AAV, and influenza A virus, have been purified using SXC. SXC is a gentle purification method with great potential for purifying large, fragile enveloped viral vectors because it does not require any chemical interaction between the target species and the stationary phase and maintains viral infectivity. The basic principles of SXC are known in the art. Briefly, a viral vector feed solution is mixed with a polyethylene glycol (PEG) buffer and loaded onto a hydrophilic stationary phase. SXC is therefore a form of filtration-assisted polyalkylene glycol precipitation. When PEG is added, depletion zones are formed around the viral particles and the stationary phase. These depletion zones are regions where the polymer center of mass cannot be reached. PEG molecules are therefore sterically excluded from these regions. The resulting depletion interaction causes the viral vector to associate with the stationary phase. Using a PEG-free buffer reverses the association between the viral particles and the stationary phase, resulting in the elution of the viral particles.

[0005] To date, SXC has only been performed on a small scale. SXC studies relying on membranes as the stationary phase have used stacked membrane layers assembled within their housings (e.g., stainless steel holders for multi-use devices or overmolded plastic housings for single-use devices), resulting in flow directed from above toward the front, resulting in dead-end flow. Previous publications of viral vector purification by SXC have used membrane devices with diameters between 13 mm and 25 mm, each with 10–20 stacked membrane layers. However, a deep mechanistic understanding of the requirements of membrane devices, especially with regard to the potential scale-up of SXC, is lacking. Current methods use typical chromatographic models that consider the entire internal surface as the interaction and binding area, leading to increased flow path, pressure, and process time upon scale-up. Neither the association location of viral vectors in the stationary phase nor the effects of various membrane device shapes or sizes on SXC performance have yet been investigated, leaving open questions about how scale-up can be achieved. Summary of the Invention [Problem to be solved by the invention]

[0006] The technical problem underlying the present invention is therefore to provide an improved means for the purification of viruses, virus-like particles, extracellular vesicles, proteins having a diameter of 5 nm or greater, protein complexes having a diameter of 5 nm or greater, nucleic acids having a diameter of 5 nm or greater, protein-nucleic acid complexes having a diameter of 5 nm or greater, and / or molecules having a diameter of 5 nm or greater using filtration-assisted polyalkylene glycol precipitation that is suitable for scale-up from medium to large production scales. [Means for solving the problem]

[0007] The solution to the above technical problem is achieved by the embodiments characterized in the claims.

[0008] In particular, in a first aspect, the present invention provides a method for purifying viruses, virus-like particles, extracellular vesicles, proteins having a diameter of 5 nm or greater, protein complexes having a diameter of 5 nm or greater, nucleic acids having a diameter of 5 nm or greater, protein-nucleic acid complexes having a diameter of 5 nm or greater, and / or molecules having a diameter of 5 nm or greater by filtration-assisted polyalkylene glycol precipitation, the method comprising: (a) providing a filtration module; Here, the filtration module is (i-α) a laminate or two separate laminates of 2 to 9 stabilized nonionic hydrophilic porous structures, the layers being placed on top of each other, wherein each of the laminate(s) exhibits a uniform flow resistance and has a total thickness of 300 μm or more and 2000 μm or less, or (i-β) one monolithic stabilized non-ionic hydrophilic porous structure or two separate monolithic stabilized non-ionic hydrophilic porous structures, each of said porous structure(s) exhibiting a uniform flow resistance and having a total thickness of 300 μm or more and 2000 μm or less; (ii) a housing containing the stack(s) of porous structure layers or the monolithic porous structure(s), the housing comprising a fluid inlet and a fluid outlet, wherein fluid entering the filtration module through the fluid inlet must pass through the stack(s) of porous structure layers or the monolithic porous structure(s) and exit the filtration module through the fluid outlet, and the housing is adapted to uniformly distribute a feed solution entering the housing via the fluid inlet over the entire accessible surface area of ​​a first layer of the stack(s) of porous structure layers or the entire accessible surface area of ​​the monolithic porous structure(s); equipped with; (b) introducing a feed solution containing the virus, virus-like particle, extracellular vesicle, protein having a diameter of 5 nm or more, protein complex having a diameter of 5 nm or more, nucleic acid having a diameter of 5 nm or more, protein-nucleic acid complex having a diameter of 5 nm or more, and / or molecule having a diameter of 5 nm or more through the fluid inlet into the first layer of the stack(s) of layers of the porous structure or the monolithic porous structure(s). 2 At least 1.0 mL per minute (1.0 mL min -1 ·cm -2 ) into the filtration module at a surface area dependent flow rate of 5 nm or more, thereby attaching viruses, virus-like particles, extracellular vesicles, proteins having a diameter of 5 nm or more, protein complexes having a diameter of 5 nm or more, nucleic acids having a diameter of 5 nm or more, protein-nucleic acid complexes having a diameter of 5 nm or more, and / or molecules having a diameter of 5 nm or more to the porous structure; (c) eluting viruses, virus-like particles, extracellular vesicles, proteins having a diameter of 5 nm or more, protein complexes having a diameter of 5 nm or more, nucleic acids having a diameter of 5 nm or more, protein-nucleic acid complexes having a diameter of 5 nm or more, and / or molecules having a diameter of 5 nm or more from the porous structure, thereby recovering purified viruses, virus-like particles, extracellular vesicles, proteins having a diameter of 5 nm or more, protein complexes having a diameter of 5 nm or more, nucleic acids having a diameter of 5 nm or more, protein-nucleic acid complexes having a diameter of 5 nm or more, and / or molecules having a diameter of 5 nm or more; The present invention relates to a method, including:

[0009] According to step (a) of the method of the present invention, there is provided a filtration module as defined above, which comprises, inter alia, (i) one stack or two separate stacks of 2 to 9, preferably 2 to 8, more preferably 3 to 7, more preferably 3 to 6, more preferably 3 to 5, layers of stabilized nonionic hydrophilic porous structures, or (ii) one or two separate monolithic stabilized nonionic hydrophilic porous structures. Thus, the filtration module may comprise 2, 3, 4, 5, 6, 7, 8, or 9 layers of stabilized nonionic hydrophilic porous structures, with 3, 4, or 5 layers of stabilized nonionic hydrophilic porous structures being particularly preferred. The layers form a stack in which the layers are placed on top of each other, preferably with maximum overlap. In this context, each stack of porous structure layers is either (i) a stack of individual porous structure layers, or (ii) a continuous porous structure layer wound on a spiral holder to form the layers and / or wound around itself to form the layers. Overall, each stack of porous structure layers exhibits a uniform flow resistance and has a total thickness of 300 μm or more and 2000 μm or less. Furthermore, when one or two separate monolithic stabilized nonionic hydrophilic porous structures are used, each exhibits a uniform flow resistance and has a total thickness of 300 μm or more and 2000 μm or less. In this context, the use of a stack of stabilized nonionic hydrophilic porous structure layers is preferred, since this results in improved uniformity of fluid flow.

[0010] The porous structure used in the present invention may have a pore size of 1 μm to 5 μm, preferably 2 μm to 4 μm, more preferably 2 μm to 3 μm, and more preferably 2.5 μm to 3 μm. Furthermore, each of the porous structure layers may have a thickness of 100 μm to 300 μm, preferably 150 μm to 300 μm, more preferably 180 μm to 270 μm, more preferably 200 μm to 250 μm, and more preferably 210 μm to 240 μm, for example, 220 μm or 230 μm.

[0011] Furthermore, when used on a laboratory scale, at least the top layer of the stack (or stacks) of porous structure layers, i.e., the first layer as viewed from upstream to downstream, preferably all of the porous structure layers as well as the monolithic porous structure, may have an accessible diameter of more than 25 mm, preferably at least 26 mm, i.e., the diameter of the largest area that can come into contact with the fluid (e.g., feed solution) supplied to the filtration module, with a material having an accessible diameter of 50 mm being particularly preferred. Thus, the layer or monolith may have a diameter of 4.91 cm 2 More than, preferably at least 5.31 cm 2 19.6 cm 2 Particularly preferred are materials with an accessible surface area of ​​10 mm to 293 mm. In other embodiments, the first layer, preferably the entire porous structure layer as well as the monolithic porous structure, viewed from upstream to downstream, can have an accessible diameter of 10 mm to 293 mm, preferably 10 mm to 60 mm, i.e., the diameter of the largest area that can contact the fluid (e.g., feed solution) supplied to the filtration module. The use of axial modules is limited to a diameter of 293 mm. Larger diameters are technically unfeasible in such module designs due to the increased pressure associated with increased membrane area. Larger diameters result in excessive pressure. Furthermore, flow distribution becomes difficult at such large diameters, making capsule or cassette designs preferable for scale-up.

[0012] For the capsule design, the incident flow area, i.e., the accessible surface area as defined above, is 6 cm 2 From 0.65m 2 Furthermore, the cassette modules can be in the range of 0.2 m due to the design of the holder and the optimized cassette for each cassette module. 2 Up to 11m from 2 For example, to purify 200 L of LV feed solution of the type described in the Methods section below, an entrance flow area of ​​3.84 m2 is required (which corresponds to 19 standard non-optimized Sartobind cassettes).

[0013] The types of porous structures that can be used in the context of the present invention are not particularly limited, as long as they have a non-ionic hydrophilic surface. In a preferred embodiment, the porous structure is a membrane or monolith, with membranes being preferred. Any membrane or monolith structure having a hydrophilic (e.g., modified) surface, such as polyethersulfone, polysulfone, polyvinylidene fluoride (PVDF), or polyamide, can be used. Respective membranes and monoliths are known in the art. Preferred membranes in this regard include stabilized cellulose membranes known in the art (e.g., Hydrosart™ membranes from Sartorius). When a monolithic porous structure is used, its total thickness is preferably 300 μm or more and 500 μm or less.

[0014] The filtration modules used in the methods of the present invention further comprise a housing as defined above, each housing being designed such that the filtration modules used in the methods of the present invention are in the form of an axial flow module, a capsule module, or a cassette module, as known in the art.

[0015] Thus, in a specific embodiment in which the filtration module is in the form of an axial flow module, the housing as defined above is a housing containing one stack of porous structure layers, the housing comprising an upper lid portion including a fluid inlet and a lower table portion including a fluid outlet, wherein the lid portion and the table portion are firmly connected to each other in a fluid-tight and pressure-tight manner, and the accessible dimensions of the stack of porous structure layers correspond to the internal dimensions of the housing, so that a fluid entering the axial flow module through the fluid inlet must pass through the stack of porous structure layers and exit the axial flow module through the fluid outlet, wherein the overall direction of fluid flow through the axial flow module is perpendicular to the stack of porous structure layers. In a preferred embodiment in this regard, the inner surface of the upper lid portion of the housing and / or the inner surface of the lower table portion of the housing, preferably both, comprise radial distribution channels that uniformly distribute the feed solution over the accessible surface area of ​​the stack of porous structure layers, and circular distribution channels that collect the fluid at the fluid outlet. In a specific embodiment, the inner surface of the upper lid portion of the housing and / or the inner surface of the lower table portion of the housing, preferably both, are provided with 8 radial distribution channels and 20 circular distribution channels. In a preferred embodiment, the stack of porous structure layers and the housing are circular in shape. Furthermore, in a preferred embodiment, the axial flow module is provided in a multi-use stainless steel holder or a single-use overmolded device, the latter being preferred. In this context, the term "the accessible dimensions of the stack of porous structure layers correspond to the internal dimensions of the housing" as used herein refers to a situation in which the inner cavity of the housing in the planar direction of the porous structure layers is completely occupied by the stack of porous structure layers. This includes, for example, a configuration in which the stack of porous structure layers has a diameter larger than the internal dimensions of the housing, but the accessible diameter of the stack becomes smaller due to a clamping edge after the housing is overmolded.

[0016] In this regard, in a specific embodiment, the method of the present invention is a method for purifying viruses, virus-like particles, extracellular vesicles, proteins having a diameter of 5 nm or greater, protein complexes having a diameter of 5 nm or greater, nucleic acids having a diameter of 5 nm or greater, protein-nucleic acid complexes having a diameter of 5 nm or greater, and / or molecules having a diameter of 5 nm or greater by filtration-assisted polyalkylene glycol precipitation, comprising: (a) providing an axial flow module; Here, the axial flow module is (i) a stack of 2 to 9 stabilized nonionic hydrophilic porous structures, the layers being placed on top of each other, wherein the stack exhibits a uniform flow resistance and has a total thickness of 300 μm or more and 2000 μm or less; (ii) a housing containing the stack of porous structure layers, the housing comprising an upper lid portion including a fluid inlet and a lower table portion including a fluid outlet, wherein the lid portion and the table portion are firmly connected to each other in a fluid-tight and pressure-tight manner, and the accessible dimensions of the stack of porous structure layers correspond to the internal dimensions of the housing, so that a fluid entering the axial flow module through the fluid inlet must pass through the stack of porous structure layers and exit the axial flow module through the fluid outlet, and the overall direction of fluid flow through the axial flow module is perpendicular to the stack of porous structure layers; equipped with; (b) introducing a feed solution containing the virus, virus-like particle, extracellular vesicle, protein having a diameter of 5 nm or more, protein complex having a diameter of 5 nm or more, nucleic acid having a diameter of 5 nm or more, protein-nucleic acid complex having a diameter of 5 nm or more, and / or molecule having a diameter of 5 nm or more through the fluid inlet into an accessible surface area of ​​1 cm of a first layer of the stack of layers of the porous structure; 2 At least 1.0 mL per minute (1.0 min -1 ·cm -2) into the axial flow module at a surface area dependent flow rate of 5 nm or more, thereby attaching viruses, virus-like particles, extracellular vesicles, proteins having a diameter of 5 nm or more, protein complexes having a diameter of 5 nm or more, nucleic acids having a diameter of 5 nm or more, protein-nucleic acid complexes having a diameter of 5 nm or more, and / or molecules having a diameter of 5 nm or more to the porous structure; (c) eluting viruses, virus-like particles, extracellular vesicles, proteins having a diameter of 5 nm or more, protein complexes having a diameter of 5 nm or more, nucleic acids having a diameter of 5 nm or more, protein-nucleic acid complexes having a diameter of 5 nm or more, and / or molecules having a diameter of 5 nm or more from the porous structure, thereby recovering purified viruses, virus-like particles, extracellular vesicles, proteins having a diameter of 5 nm or more, protein complexes having a diameter of 5 nm or more, nucleic acids having a diameter of 5 nm or more, protein-nucleic acid complexes having a diameter of 5 nm or more, and / or molecules having a diameter of 5 nm or more; The method includes:

[0017] In another specific embodiment in which the filtration module is in the form of a capsule module, the housing as defined above is a housing containing a stack of porous structure layers, the housing having a fluid inlet through which the fluid flows into the outer fluid channel, and from which the fluid passes through the stack of porous structure layers to the inner fluid channel connected to the fluid outlet, where the stack of porous structure layers is firmly connected to the outer and inner flow channels in a fluid-tight and pressure-tight manner, and the accessible dimensions of the stack of porous structure layers correspond to the internal dimensions of the housing, so that the fluid entering the capsule module through the fluid inlet must pass through the stack of porous structure layers and exit the capsule module through the fluid outlet, where the overall direction of fluid flow through the capsule module is radial to the stack of porous structure layers. In such capsule modules, the stack of porous structure layers is preferably wound, or the capsule has a porous structure layer incorporated by pleating technology. In these capsule modules, the feed flows radially from the outer fluid channel through the porous structure layers to the inner fluid channel. Furthermore, the porous structure layer can be a porous structure layer wrapped around the inner fluid channel, a pleated porous structure layer, or a folded porous structure layer. Another possible module design is a spiral-wound module. In the case of a wrapped porous structure layer, a stack of 2 to 9 stabilized nonionic hydrophilic porous structures is provided, where the layers are wrapped around the inner fluid channel, where the stack exhibits uniform flow resistance and has a total thickness of 300 μm or more and 2000 μm or less. A fleece material may additionally be incorporated. Pleated, folded, or other porous structure arrangements are possible, which should result in a thickness similar to that of the wrapped type.

[0018] In another specific embodiment in which the filtration module is in the form of a cassette module, the housing as defined above is a housing containing two separate stacks of porous structure layers or two separate monolithic stabilized non-ionic hydrophilic porous structures, the housing having a fluid inlet from which the fluid enters into a central inlet flow channel, where the fluid enters at the top between the two stacks or the two monolithic porous structures, passes through the stack of porous structure layers or the monolithic porous structure, and heads towards an external (downstream) channel connected to a fluid outlet, where the two stacks of porous structure layers or the monolithic porous structure The two monolithic porous structures are firmly connected to the central inlet flow channel and the external channel in a fluid-tight and compact manner, and the accessible dimensions of the stack of porous structure layers or the monolithic porous structure correspond to the internal dimensions of the housing, so that the fluid entering the cassette module through the fluid inlet must pass through the two stacks of porous structure layers or the monolithic porous structure and exit the cassette module through the fluid outlet, where the overall direction of fluid flow through the cassette module is perpendicular to the stack of porous structure layers or the monolithic porous structure.

[0019] The current cassette module design features a stack of 15 membranes for membrane adsorption chromatography, resulting in a bed height of 4 mm. The filtration area is 220 mm x 470 mm. The height of the channel distributing the pre-filtration sample is over 5 mm, and the total height of the filtrate channel is over 4 mm, resulting in a total height of 0.2 m. 2This results in a total dead volume of 1600 mL for a total filtration area of ​​100 mL. Unlike chromatographic requirements, the PEG filtration device according to the present invention in the form of the cassette module described above uses 2 to 9, preferably 3 to 5, membrane layers, resulting in a thickness of 0.5 to 1 mm. To achieve the same flux level across the entire area, two or more layers are required. For the intended process, it is preferable that the entire filtration area be simultaneously addressed. To achieve this, dead volume optimization and flow optimization are preferred. In particular, not only can the thickness of the filter stack(s) ((2) in Figure 8) be reduced from 4 mm to 1 mm, but the upstream distribution plate ((1) in Figure 8) can be reduced to a channel height of less than 2.5 mm, and the filtrate spacer ((3) in Figure 8) can be reduced to less than 2.5 mm.

[0020] In step (b) of the method of the present invention, the feed solution is introduced through the fluid inlet into the first layer(s) of the stack(s) of layers of the porous structure or into the monolithic porous structure(s) over an accessible surface area of ​​1 cm.sup.2. 2 At least 1.0 mL per minute (1.0 mL min -1 ·cm -2 ), preferably with an accessible surface area of ​​1 cm 2 At least 1.1 mL per minute per accessible surface area, more preferably 1 cm 2 At least 1.2 mL per minute per accessible surface area, more preferably 1 cm 2 At least 1.3 mL per minute per accessible surface area, more preferably 1 cm 2 The flow module is supplied with a surface-area-dependent flow rate of at least 1.4 mL per minute per cm of accessible surface area. Of note, there is no general specific upper limit for the surface-area-dependent flow rate. However, technical limitations (e.g., regarding maximum pressure) lead to effective upper limits depending on the specifications of the porous structure and the maximum pressure of the housing module. In a specific embodiment, the surface-area-dependent flow rate is at least 1.4 mL per cm of accessible surface area. 2 1.0 mL to 3.5 mL per minute (1.0 mL min -1 ·cm -2 ~3.5mL min-1 ·cm -2 ), preferably with an accessible surface area of ​​1 cm 2 1.1 mL to 2.2 mL per minute (1.1 mL min -1 ·cm -2 ~2.2mL min -1 ·cm -2 ), more preferably an accessible surface area of ​​1 cm 2 1.1 mL to 1.6 mL per minute (1.1 mL min -1 ·cm -2 ~1.6 mL min -1 ·cm -2 In a specific embodiment, the range is 1.426 mL min -1 ·cm -2 In this context, the surface area dependent flow rates are based not on the total accessible area, but on the accessible surface area of ​​the first (i.e., the top / most upstream layer, as viewed from upstream to downstream) layer of a stack(s) of porous structure layers, or of the monolithic porous structure(s). Thus, by way of example, a surface area dependent flow rate of, for example, 5.31 cm 2 A stack of two porous structure layers each having an accessible surface area of ​​5.31 cm 2 Both stacks of nine porous structure layers have an accessible surface area of ​​5.31 cm , and the top membrane of the membrane stack has the same accessible surface area of ​​5.31 cm . 2 has an accessible surface area of

[0021] In a preferred embodiment, prior to step (b) of the method of the present invention, the feed solution and buffer are mixed in a buffer:feed solution ratio of 4:1 to 1:4, preferably 2:1 to 1:2, more preferably 1.5:1 to 1:1.5, more preferably 1.2:1 to 1:1.2. In a specific embodiment, the feed solution containing the virus or virus-like particles and the buffer are mixed in a buffer:feed solution ratio of 1:1. The mixing is preferably performed by dynamic in-line mixing.

[0022] The respective buffers used in this regard can be selected from buffers that provide an optimal environment for purified viruses, virus-like particles, extracellular vesicles, proteins having a diameter of 5 nm or more, protein complexes having a diameter of 5 nm or more, nucleic acids having a diameter of 5 nm or more, protein-nucleic acid complexes having a diameter of 5 nm or more, and / or molecules having a diameter of 5 nm or more, and such buffers are known in the art. Notably, the respective buffer compositions can be advantageously used throughout the methods of the present invention. In a preferred embodiment, the buffer comprises a polyalkylene glycol having 2 or 3 carbon atoms in the repeating unit, where the polyalkylene glycol is preferably polyethylene glycol (PEG), preferably PEG having a molar mass between 400 g / mol and 12,000 g / mol, preferably between 2,000 g / mol and 12,000 g / mol, more preferably between 3,000 g / mol and 8,000 g / mol. In a specific embodiment, PEG having a molar mass of 4,000 g / mol (PEG 4000) is used. Preferably, the buffer solution contains polyalkylene glycol or PEG in an amount of 4% to 20% by weight, preferably 5% to 17.5% by weight, more preferably 7.5% to 15% by weight, and more preferably 10% to 15% by weight, resulting in a final concentration of polyalkylene glycol or PEG after mixing with the feed solution. In certain embodiments, the buffer solution contains polyalkylene glycol or PEG in an amount of 25% by weight, and the buffer:feed solution mixing ratio is 1:1, so that the amount of polyalkylene glycol or PEG in the feed solution after mixing of the buffer solution and the feed solution is 12.5% ​​by weight. Other PEG sizes can be used, with concentrations adapted to reflect their molecular weights. The higher the molecular weight of the PEG, the lower the required PEG concentration.

[0023] According to the present invention, in step (b) of the method of the present invention, the feed solution is fed to the filtration module in a downflow direction, ie from the fluid inlet towards the fluid outlet.

[0024] In the elution step (c) of the method of the present invention, an elution buffer containing no polyalkylene glycol or PEG is used, or an elution buffer containing a lower amount of either polyalkylene glycol or PEG than that used in the buffer used for mixing before step (b). In this context, a lower amount is preferably 80% or less, more preferably 50% or less, more preferably 20% or less, and most preferably 0% of the concentration of polyalkylene glycol or PEG in the buffer used for mixing before step (b). The higher the concentration of polyalkylene glycol or PEG in the elution buffer, the broader the elution peak, resulting in a lower concentration of the target product in the elution. Of note, gradient elution, as known in the art, can be used in this step, in which the amount of either polyalkylene glycol or PEG is gradually reduced from the amount of either polyalkylene glycol or PEG used in the buffer used for mixing before step (b) to a buffer containing no polyalkylene glycol or PEG. Nevertheless, preferably, the elution in step (c) of the method of the present invention uses an elution buffer that does not contain any polyalkylene glycol or PEG. Thus, in a specific embodiment, the elution buffer is the same as the buffer used for mixing prior to step (b), but does not contain any polyalkylene glycol or PEG. Elution may be carried out in either a downflow or an upflow direction, i.e., from the fluid outlet (which in this case effectively functions as a fluid inlet) to the fluid inlet (which in this case effectively functions as a fluid outlet), with the latter case being preferred.

[0025] As noted above, the method of the present invention is based on the SXC virus purification process known in the art. Thus, the method of the present invention may include an equilibration step prior to step (b), as well as one or more washing steps after equilibration and before step (b), and / or after step (b) and before step (c).

[0026] The viruses that can be purified in the method of the present invention are not particularly limited, and include large enveloped viruses known in the art. Preferably, the viruses purified in the method of the present invention are selected from the group consisting of lentivirus, baculovirus, orf virus, adeno-associated virus (AAV), influenza virus (such as influenza A virus), yellow fever virus, human papillomavirus, vaccinia virus, adenovirus, hepatitis virus, poliovirus, rabies virus, rotavirus, rubella virus, and Zika virus, and their respective virus-like particles. In a specific embodiment, the virus is a lentivirus. In another specific embodiment, the virus is an AAV.

[0027] In this context, as used herein, the terms "virus" and "viral vector" are used interchangeably.

[0028] Furthermore, the virus-like particles that can be purified by the method of the present invention are not particularly limited, and include nucleic acid-free virus particles, including viroids, virusoids, and prions.

[0029] Furthermore, the present invention can purify extracellular vesicles (such as exosomes or microvesicles), lipid nanoparticles, proteins or protein complexes, nucleic acids, and protein-nucleic acid complexes, each having a diameter of 5 nm or more. Potential protein targets can be antibodies, antibody-drug conjugates (ADCs), bispecific or multispecific molecules, membrane proteins, and membrane protein complexes, especially large membrane protein complexes. Preferably, the proteins and molecules that can be purified by the present invention are globular proteins and globular molecules, respectively.

[0030] As described herein, the method of the present invention is filtration-assisted polyalkylene glycol precipitation method.However, in related embodiments, the precipitation in this method can also be carried out by ammonium sulfate precipitation, precipitation using polymers other than polyalkylene glycol, polyelectrolyte precipitation or pH shift precipitation, as known in the art.

[0031] In a second aspect, the present invention provides a filtration module comprising: (a-α) a laminate or two separate laminates of 2 to 9 stabilized nonionic hydrophilic porous structures, the layers being placed on top of each other, wherein each of the laminate(s) exhibits a uniform flow resistance and has a total thickness of 300 μm or more and 2000 μm or less, or (a-β) one monolithic stabilized non-ionic hydrophilic porous structure or two separate monolithic stabilized non-ionic hydrophilic porous structures, each of said porous structure(s) exhibiting a uniform flow resistance and having a total thickness of 300 μm or more and 2000 μm or less; (b) a housing containing the stack(s) of porous structure layers, the housing having a fluid inlet and a fluid outlet, wherein fluid entering the filtration module through the fluid inlet must pass through the stack(s) of porous structure layers and exit the filtration module through the fluid outlet, and the housing is adapted to uniformly distribute a feed solution entering the housing via the fluid inlet over the entire accessible surface area of ​​a first layer of the stack(s) of porous structure layers; The present invention relates to a filtration module comprising:

[0032] A flow module according to the invention is as defined in the above first aspect of the invention.

[0033] In particular, all provisions of the flow module provided in the context of the inventive method according to the first aspect of the invention apply equally to the flow module according to the second aspect of the invention itself.

[0034] Thus, in a specific embodiment, the flow module of the present invention is an axial flow module comprising: (a) a stack of 2 to 9 stabilized nonionic hydrophilic porous structures, the layers being placed on top of each other, wherein the stack exhibits a uniform flow resistance and has a total thickness of 300 μm or more and 2000 μm or less; (b) a housing containing the stack of porous structure layers, the housing comprising an upper lid portion including a fluid inlet and a lower table portion including a fluid outlet, wherein the lid portion and the table portion are firmly connected to each other in a fluid-tight and pressure-tight manner, and the accessible dimensions of the stack of porous structure layers correspond to the internal dimensions of the housing, so that a fluid entering the axial flow module through the fluid inlet must pass through the stack of porous structure layers and exit the axial flow module through the fluid outlet, and wherein the overall direction of fluid flow through the axial flow module is perpendicular to the stack of porous structure layers; The axial flow module includes:

[0035] In such embodiments, the interior surface of the upper lid portion of the housing and / or the interior surface of the lower table portion of the housing, preferably both, preferably comprises radial distribution channels that uniformly distribute the virus or virus-like particle containing feed solution over the accessible membrane surface area of ​​the membrane stack, and circular distribution channels that collect the fluid towards the fluid outlet. In a specific embodiment, the interior surface of the upper lid portion of the housing and / or the interior surface of the lower table portion of the housing, preferably both, comprise 8 radial distribution channels and 20 circular distribution channels.

[0036] In another specific embodiment, the flow module of the present invention is a capsule module comprising: (a) a stack of 2 to 9 stabilized nonionic hydrophilic porous structures, the layers being placed on top of each other, wherein the stack exhibits a uniform flow resistance and has a total thickness of 300 μm or more and 2000 μm or less; (b) a housing that accommodates the stack of porous structure layers, the housing having a fluid inlet through which fluid flows into an outer fluid channel, and from which fluid passes through the stack of porous structure layers to an inner fluid channel connected to a fluid outlet, wherein the stack of porous structure layers is firmly connected to the outer flow channel and the inner flow channel in a fluid-tight and pressure-tight manner, and the accessible dimensions of the stack of porous structure layers correspond to the internal dimensions of the housing, so that fluid entering the capsule module through the fluid inlet must pass through the stack of porous structure layers and exit the capsule module through the fluid outlet, and wherein the overall direction of fluid flow through the capsule module is radial to the stack of porous structure layers; The capsule module includes:

[0037] In another specific embodiment, the flow module of the present invention is a cassette module comprising: (a-α) a laminate or two separate laminates of 2 to 9 stabilized nonionic hydrophilic porous structures, the layers being placed on top of each other, wherein each of the laminate(s) exhibits a uniform flow resistance and has a total thickness of 300 μm or more and 2000 μm or less, or (a-β) one monolithic stabilized non-ionic hydrophilic porous structure or two separate monolithic stabilized non-ionic hydrophilic porous structures, each of said porous structure(s) exhibiting a uniform flow resistance and having a total thickness of 300 μm or more and 2000 μm or less; (b) A housing containing two separate stacks of the porous structure layers or two separate monolithic stabilized non-ionic hydrophilic porous structures, the housing having a fluid inlet through which fluid flows into a central inlet flow channel, where the fluid enters the upper portion between the two stacks or the two monolithic porous structures, passes through the stack of porous structure layers or the monolithic porous structures, and heads toward an external (downstream) channel connected to a fluid outlet, where the two stacks of porous structure layers or the two monolithic porous structures are connected to the central inlet flow channel. a housing, the housing being firmly connected to the inner channel and the outer channel in a fluid-tight and pressure-tight manner, and the accessible dimensions of the stack of porous structure layers or the monolithic porous structure correspond to the internal dimensions of the housing, so that a fluid entering the cassette module through the fluid inlet must pass through the two stacks of porous structure layers or the monolithic porous structure and exit the cassette module through the fluid outlet, wherein the overall direction of fluid flow through the cassette module is perpendicular to the stack of porous structure layers or the monolithic porous structure; A cassette module comprising:

[0038] As used herein, the term "comprising" / "comprises" expressly includes the terms "consisting essentially of" / "consists essentially of" and "consisting of" / "consists of", i.e., all such terms are interchangeable herein.

[0039] In this study, we demonstrated the location of LV and AAV on the stationary phase of a small-scale device. Based on these results, we investigated an intermediate scale-up approach using a device with four times the membrane surface area. This not only highlighted the key aspects of flow rate scaling, but also the importance of modular design for successful viral vector recovery using filtration-assisted polyalkylene glycol precipitation in a scale-up format.

[0040] In this context, it is noted that when considering SXC as a chromatographic method, it is assumed that the entire column is used at depth, making scalability difficult. However, the present invention has been shown to advantageously avoid the use of depth. Thus, the present invention refers to filtration-assisted polyalkylene glycol precipitation. This differs from commonly known SXC processes, which use the entire depth of the stationary phase, not just the membrane layer up to a depth of about 300 μm.

[0041] Specifically, we addressed the scaling-up of lentiviral vector purification by SXC using an axial-flow membrane module. Visualization of LVs captured on the membrane during SXC revealed that this is a surface-directed process that primarily utilizes the upper membrane layer. Transferring the SXC process to a module with a fourfold larger area revealed that scaling the flow rate was a critical factor and must be related to the membrane area of ​​the first (most upstream) membrane layer. With optimized conditions and an overmolded membrane device, we achieved a 73% recovery of infectious lentiviral vector and 77% and 63% protein and dsDNA removal rates. We constructed a prototype housing for the module and analyzed LV capture on the membrane and SXC performance. Our findings demonstrated that not only the flow rate relative to the membrane area of ​​the first membrane layer but also uniform membrane utilization are important in this regard.

[0042] Thus, the present invention advantageously recognizes that virus or virus-like particle adsorption to stabilized cellulose membranes during SXC is a primarily surface-dependent process, leading to the surprising discovery that any scale-up of SXC for virus or virus-like particle purification requires not only scaling up the flow rate relative to the accessible membrane area of ​​the top membrane in the membrane stack, but also optimizing fluid distribution to the membrane.

[0043] In modular designs, the surface-related capture process necessitates larger membrane surface areas as capacity demands increase, so scaling up results in proportionally higher flow rates over larger surface areas, while the transmembrane pressure as well as the processing time remains constant. [Brief explanation of the drawings]

[0044] [Figure 1]SXC membrane device. (A) MA15 (25 mm membrane diameter) in a stainless steel holder or (B) overmolded. (C) Internal structure of the lid and table of the MA15 with radial and circular distribution channels, and (D) cross-section of the SXC membrane device. (E) MA100 (50 mm membrane diameter) in a stainless steel holder or (F) overmolded. (G) Internal support geometry of the lid of the MA100, and (H) structure of the table of the MA100 with radial and circular distribution channels. [Figure 2] Figure 1 shows the membrane layers of an MA15 device after loading with lentiviral vectors labeled with anti-VSV-G Alexa Fluor™ 546 antibody. (A) The membrane layers were separated after the loading and washing steps of steric exclusion chromatography, or (B) after the elution step. A membrane not incorporated into the device served as a negative control. [Figure 3] Figure 1 shows the identification of process parameters critical for achieving high infectious LV recovery. (A) Infectious LV recovery at various flow rates between 3 mL min and 9 mL min. (B) Infectious LV recovery for MA15 and MA100 at the same surface-area-dependent flow rate of 1.426 mL min cm. MA15 (N = 6), MA100 (N = 11). (C) Infectious LV recovery for internally mixed and externally mixed LV-PEG solutions (N = 3). Data in B and C represent the mean ± standard deviation. p values ​​are indicated as follows: *p ≤ 0.05, **p ≤ 0.01, ns not significant. [Figure 4]Figure 1 shows impurity removal by steric exclusion chromatography. (A) Protein and dsDNA removal rates using MA15 and MA100 devices at the same surface-area-dependent flow rate of 1.426 mL min cm. (N = 6 for MA15 and N = 11 for MA100). (B) Silver-stained SDS-PAGE gel of SXC fractions: 1 - marker, 2 - loading material, 3 - flow-through, 4 - wash, and 5 - eluate. Protein bands represent VSV-G envelope proteins, reverse transcriptase (RT) subunit p51 and subunit p66, integrase (INT), capsid (CA), and matrix (MA). Data in A represent the mean ± standard deviation. [Figure 5] Figure 1. Determination of loading volume. Infectious LV recovery when loading (A) 200 mL or (D) 700 mL. Phase contrast images merged with green fluorescence channel images of HEK293T cells after incubation with (B) the flow-through or (C) the elution fraction. Main effect plots of infectious (E) LV recovery and (F) LV particle recovery for various loading volumes. [Figure 6] Figure 1 shows the membrane layers of various MA100 device designs after loading with lentiviral vectors labeled with anti-VSV-G Alexa Fluor™ 546 antibodies. The membrane layers were separated after the loading and washing steps of steric exclusion chromatography. The configurations of the housing lid and table are shown on the left: (A) standard configuration, (B) inverted configuration, and (C) prototype configuration. [Figure 7] Figure 1 shows LV recovery for various MA100 housing configurations. Infectious and particulate LV recovery for (A) standard and inverted configurations and (B) standard and prototype configurations of MA100 housing operated at 1.426 mL min cm (N=3 each). Data represent mean ± standard deviation. [Figure 8]Figure 1 shows a standard cassette design. Standard cassette design of an existing cassette module. The module consists of only a distribution plate 1, membrane stack(s) 2 and filtrate spacers 3. Based on this standard configuration, an optimized cassette design according to the present invention is conceived. [Figure 9] Surface area specific flow rate. (A) Particulate LV recovery and (B) infectious LV recovery, as well as (C) dsDNA removal and (D) protein removal rates for SXC experiments plotted against various surface area specific flow rates for various module sizes. The number of replicates for each device and flow rate is as follows: PP15 (N = 3 for all flow rates), MA15 (N = 3 for all flow rates), MA100 (N = 11 for 1.43 mL min cm, N = 3 for all other flow rates), 5 inch capsule (N = 1). [Figure 10] This figure shows the membrane layers of an MA15 device after loading adeno-associated vectors labeled with Capture select Alexa Fluor 647 anti-AAVX conjugate. AAV8 lysate was incubated with the antibody (1:3000) and incubated overnight at 4°C. Membrane layers 1 to 5 were separated after the loading and washing steps of steric exclusion chromatography (top row) or after the elution step (bottom row). [Figure 11] Figure 1 shows AAV8 recovery in various steric exclusion chromatography fractions using 15% PEG 4000. A final 15% PEG 4000 concentration was achieved by in-line mixing of the PEG buffer with the AAV solution. The AAV8 lysate was loaded onto 15 MA units. [Figure 12] Figure 1 shows a chromatogram of a representative SXC run in which AAV was purified using 15% PEG 4000. A final 15% PEG 4000 concentration was achieved by in-line mixing of the PEG buffer with the AAV solution. AAV8 lysate was loaded onto 15 units of MA. [Figure 13]Figure 1 shows AAV recovery in various steric exclusion chromatography fractions using 10% PEG 4000. A final 10% PEG 4000 concentration was achieved by in-line mixing of the PEG buffer with the AAV solution. AAV8 lysate was loaded onto 15 MA units. [Figure 14] Figure 1 shows a chromatogram of a representative SXC run purifying AAV using 10% PEG. A final 15% PEG 4000 concentration was achieved by in-line mixing of the PEG buffer with the AAV solution. AAV8 lysate was loaded onto 15 units of MA. DETAILED DESCRIPTION OF THE INVENTION

[0045] The present invention will be further illustrated by the following examples, but the present invention is not limited thereto. [Example]

[0046] Materials and Methods: Lentiviral vector generation, harvest, and clarification Third-generation lentiviral vectors were produced by transient transfection of four plasmids into suspension HEK293T / 17 SF cells (ACS-4500, ATCC) in a UniVessel™ 10L bioreactor operated by a BIOSTAT™ B-DCU (Sartorius). The pH electrode was calibrated, and the vessel was assembled (accommodating a 2 x 3-blade segment impeller and a ring-up sparger) and filled to 30% of its volume with water. The bioreactor was autoclaved at 121°C. After autoclaving, the bioreactor was emptied and filled to 80% of its final volume with FreeStyle medium (Thermo Fisher Scientific) + 0.0002% Antifoam C (Sigma-Aldrich) + 1x insulin-transferrin-selenium (Thermo Fisher Scientific). The bioreactor was connected to the BIOSTAT™, the DO probe was calibrated, and the pH electrode was recalibrated. The culture settings were 202 rpm, 30% DO, 37°C, and pH 7.1. The bioreactor was left overnight to adjust the pH and pO2. The next day, 0.3 x 10 per mL of ethanol was added. 6The bioreactor was inoculated with 9% of the final volume of the bioreactor at a final viable cell density of 1000 cells. After inoculation, samples were taken from the bioreactor daily and thereafter, and viable cell density and viability were determined using a Cedex HiRes (Roche) and offline pH measurement. If a difference of more than 0.1 was detected between the externally and internally measured pH, the pH probe was recalibrated. Transfection was performed 3 days after inoculation. 0.5 mg of total plasmid DNA per liter of final culture volume was used in a mass ratio of 5:2.5:1:1 (pALD-Lenti-GFP:pALD-GagPol:pALD-VSV-G:pALD-REV1; Aldevron) prepared in additive-free FreeStyle medium. In a separate flask, 4 mL of PEIpro per 1 mg of total plasmid DNA was diluted in FreeStyle (each at 5% of the final bioreactor volume). The two solutions were mixed and incubated for 15 minutes before being added to the bioreactor. Eighteen hours after transfection, the following reagents were added to the bioreactor: anti-aggregation agent (1:500 (vol / vol)), the enhancer sodium butyrate (10 mM final concentration), and 1 mL of 2% Antifoam C. Nuclease treatment to digest nucleic acids was performed at 10 U mL. -1 The digestion was carried out directly in the bioreactor at 37°C for 1 hour using 1000µL of DENARASE™ (c-Lecta) and 2mM MgCl2 (final concentration). After nucleic acid digestion, the cell culture broth (containing the lentiviral vector) was clarified using a Sartoclear Dynamics™ Lab V50 (0.45µm polyethersulfone membrane) containing 5g / L diatomaceous earth (Sartorius). The lentiviral vector was aliquoted and stored at -80°C.

[0047] Membrane and Housing Stabilized cellulose membrane Hydrosart™ 10242 (Sartorius) was used as the stationary phase. The membrane lot used in this study had a thickness of 220 μm per layer and a mean flow pore size of 2.5 μm to 3 μm. A stack of five membrane layers with a diameter of 57 mm was assembled into an MA100 polypropylene module and overmolded using an Arburg 221-75-350 injection molding machine. The final chromatography module had an accessible membrane diameter of 50 mm, resulting in a total of 19.63 cm per layer. 2 This results in an accessible membrane surface area of ​​100 μm. The maximum pressure recommended for this device is 0.6 MPa. For comparison, a smaller-scale SXC device, MA15, incorporating the same membrane (5 layers), was used. Additionally, an axial PP15 device with a diameter of 16 mm and a 5-inch capsule were used for further scale-up experiments. For lentiviral vector visualization experiments, the membrane and housing were incorporated into a stainless steel holder to allow easy access to the membrane for visualization. The MA15 membrane device and MA100 membrane device are shown in Figure 1, and the specifications of all modules are listed in Table 1. The MA15 has one-quarter of the surface area and membrane volume of the MA100 device.

[0048] The MA100 membrane housing used in this study has a lid and table with different geometries. The lid has a textured structure with a thicker bridge, which is intended to prevent the membrane from being pressed tightly against the lid and allow room for the liquid to spread (Figure 1G). The table has eight radial distribution channels and 20 circular distribution channels that collect the fluid at the outlet (Figure 1H). The PP15 and MA15 devices have the same distribution channel geometry in the lid and table (Figure 1C).

[0049] [Table 1]

[0050] Chromatography equipment and procedures Lentiviral vectors were purified by SXC using the AEKTA™ avant 150 (Cytiva Life Sciences) chromatography system with in-line UV (280 nm) and conductivity monitoring (operated by UNICORN 7.1 software) and PP15, MA15, and MA100 modules. For large-scale SXC experiments using 5-inch capsules (Sartorius) with a 4 mm bed height, a multi-use membrane chromatography system (MU RCC, Sartorius) was used, equipped with a PuraLev™ i30SU pump (Levitronix) operating in-line at 600 rpm, which served as a dynamic mixer for buffer and feed solutions. All chemicals (Tris, hydrochloric acid (HCl), sodium chloride (NaCl), PEG 4000) were purchased from Carl Roth. Buffers were prepared in ultrapure water in an Arium™ Pro (Sartorius) system. Two buffers were prepared: 1) 50 mM Tris-HCl buffer (pH 7.4) containing 150 mM NaCl (A1), and 2) 25% PEG 4000 in 50 mM Tris-HCl, 150 mM NaCl (pH 7.4) (B1), hereafter referred to as Tris buffer and PEG buffer, respectively.

[0051] Chromatographic runs with MA15 devices were performed as known in the art, and volumes for equilibration, loading, washing, and elution were adapted to MA100 devices as listed in Table 1.

[0052] On the day of the experiment, LV samples were thawed in a 37°C water bath until only small ice chunks remained. The samples were then stored at 4°C until use (30–60 min). The entire LV solution was used on the day of thawing. Different LV batches were used for different experiments. Therefore, the respective titers of each LV sample are shown in the following examples. The LV solution was kept on ice during the experiment, and fractions were collected and cooled at 4°C. First, the MA100 membrane device was equilibrated with an in-line mixture of Tris buffer and PEG buffer at a 1 / 2 dilution. Then, a PEG buffer with a concentration of 25% (wt / vol) PEG 4000 was added to reach a final PEG concentration of 12.5%. Unless otherwise indicated, the LV sample (A2) was loaded by in-line mixing with the PEG buffer at a 1 / 2 dilution in the downflow direction. The loading volume varied between experiments and is shown for each experiment in the following examples. The membrane column was washed with an in-line mixture of Tris buffer and PEG buffer at a 1 / 2 dilution. LV was eluted with Tris buffer in the upflow direction. Fractions were aliquoted and stored at -80°C for analysis. The flow rate, loading volume, and design of the SXC membrane device varied depending on the experiment and are listed in each example below. High flow rates (28 mL min -1 ), an open configuration of the chromatography system was used to reduce pressure by manually collecting fractions immediately after the chromatography device without flowing through the entire system. A new membrane device was used for each run.

[0053] analysis: Determination of infectious titer The infectious titer of LV was quantified using an Incucyte™ S3 live cell analysis system (Sartorius). Adherent HEK293T cells (ACC 635, DSMZ) were infected with serially diluted LV samples, and GFP expression was measured by real-time imaging as known in the art, with the modification that no staining was performed and transgene expression (GFP) was read 48 hours post-infection. Samples were analyzed in duplicate.

[0054] Determination of particle titer LV particle titers were quantified by enzyme-linked immunosorbent assay (ELISA) using the QuickTiter™ Lentivirus Titer Kit (Cell Biolabs) to quantify p24 capsid protein. Absorbance was read at 450 nm using a FLUOstar Omega plate reader (BMG Labtech). The resulting standard curve was fitted by a second-order polynomial. The determined p24 concentration was 1.25 × 10 7 The titer was converted to a viral particle titer by assuming that each LV particle contains 1 ng of p24 and that one LV particle contains approximately 2000 molecules of p24.

[0055] Total protein quantification Total protein concentration was determined using the Pierce™ Coomassie Bradford Protein Assay Kit (Thermo Fisher Scientific) according to the manufacturer's instructions. Standards and samples were analyzed in duplicate in a clear 96-well microtiter plate (Greiner Bio-one). Absorbance was read at 595 nm using a microplate reader. The resulting standard curve was fitted by linear regression.

[0056] Quantification of total dsDNA Total dsDNA was quantified using the Quant-iT™ Pico-Green™ dsDNA Assay (Thermo Fisher Scientific) according to the manufacturer's instructions. Standards and samples were analyzed in duplicate in a black 96-well microplate (Corning). Samples were excited at 480 nm, and fluorescence emission intensity was measured at 520 nm using a microplate reader. The resulting standard curve was fitted by linear regression.

[0057] SDS-PAGE and silver staining Proteins were fractionated by SDS-PAGE on 4%-15% Mini-PROTEAN™ TGX Stain-Free Protein Gels (Bio-Rad). SDS-PAGE was performed according to the manufacturer's instructions. Precision Plus Protein Standards (Bio-Rad) were used as markers. Gels were run at a constant voltage of 300 V for 15-20 minutes. Protein bands were visualized using a Pierce Silver Stain Kit (Thermo Fisher Scientific).

[0058] Visualization of lentiviral vectors Staining was performed to visualize the location of LVs on the membrane before and after elution. LV samples were incubated with a mouse monoclonal antibody against VSV-G (F-6) labeled with Alexa Fluor™ 546 (Santa Cruz Biotechnology) at a dilution of 1:2000 for 1 h at 4°C. Five layers of Hydrosart™ membranes were placed between the table and lid of a chromatography device assembled in a stainless steel holder (Figures 1A and 1E). The screws were tightened to 3 Nm. SXC runs were performed as described above and stopped after the pre-elution wash step. Membranes were separated and visualized using UVP ChemStudio (Analytik Jena) by applying a green light source (550 nm), an ethidium bromide filter, and a 60-second exposure time. An untreated membrane layer not assembled in the membrane holder device served as a negative control.

[0059] Statistical analysis: The statistical significance of differences between groups was assessed using an unpaired Student's t-test (two-tailed) using OriginPro™ 2021 (OriginLab). Where applicable, experiments were evaluated using MODDE Pro 13 (Sartorius).

[0060] Example 1: Visualization of lentiviral vectors on membranes So far, it is unclear where the target (in this case, LV) is located on the membrane after loading by SXC, and there is a lack of studies of particle localization on the stationary phase during SXC in the literature. Visualization of LV on the stationary phase may contribute to understanding the SXC process and process requirements.

[0061] Lentiviral vectors were stained with anti-VSV-G antibody labeled with Alexa Fluor™ 546 as described above. Labeled LVs were loaded onto a membrane integrated into an MA15 housing and placed in a stainless steel holder (Figure 1A). A final PEG 4000 concentration of 12.5% ​​and 7 mL min were used. -1 SXC was performed using PEG buffer with a flow rate of 1.5 x 10 per mL. A volume of 50 mL was loaded. 11 virus particles (VP ​​mL -1 ) corresponds to a volume of 25 mL of LV solution containing 100 μL of LV solution. SXC runs were stopped after the loading and washing steps (Figure 2A) or after the elution step (Figure 2B) and optically visualized.

[0062] Figure 2 shows the visualization of viral vectors captured on the stationary phase after loading by SXC. LV particles were primarily present on the first and second layers of the membrane. Some LV particles could be detected on layer 3, but no fluorescence was detected on layers 4 and 5. Particles were homogeneously distributed across the membrane layers. Only the clamping edges, which do not come into contact with the liquid, were correspondingly unstained. These findings indicate that very few viral particles were captured in the deeper layers of the unit with SXC. Therefore, using 15 layers, as often described in the literature, does not appear to offer any added value compared to using 5 layers. Furthermore, the column volume specified for other classical chromatography devices does not play a major role in the SXC method. Although this is a simple technique, visualization of viral vectors on the membrane points out that adding more membrane layers (and thereby increasing membrane volume) does not appear to be a useful scaling method. In SXC, the surface area of ​​the first layer is the more important feature. It is assumed that once the first layer is saturated, access to further layers is restricted, resulting in stacking of multilayered LV particles and a reduction in pore size. Therefore, as previously reported, a pressure increase is observed during loading. After elution, no fluorescence was detected on the membrane, indicating that (almost) all of the LVs were eluted.

[0063] Example 2: Identifying process parameters critical to SXC scale-up SXC has previously only been performed on a small scale using axial membrane devices with diameters up to 25 mm. By increasing the membrane surface area fourfold, we sought to identify process parameters critical for successful scale-up of lentiviral vector purification by SXC. In a previous study using a small-scale MA15 device, 12.5% ​​PEG 4000 was determined to be the ideal buffer for the purification of LV by SXC. Therefore, this buffer composition was used and not further modified in the following experiments. In the same study using the MA15, a 6 mL min -1 ~7mL min -1 (Tested flow rate range 3 mL min-1 ~9 mL min -1 ) was identified, achieving infectious LV recovery rates of over 80%. In the first attempt, the MA100 device and 1.25 × 10 per mL 7 Transducing units (TU mL -1 ) using an LV batch with a titer of 3 mL min -1 to 9 mL·min -1 Flow rates between 0.01 and 0.1 were tested. Figure 3A shows that lower than expected infectious LV recovery rates were observed. It was concluded that the optimal flow rate for the MA100 device was not within this range.

[0064] In the case of membrane chromatography, flow rates are typically expressed in terms of membrane volume per mL. As discussed in Example 1, scaling up by simply increasing the membrane volume without increasing the surface area of ​​the first membrane is unlikely to be useful for surface-directed capture of vector particles. Therefore, it is more reasonable to specify the flow rate per membrane surface area (of one layer) rather than per membrane volume. Considering the dynamic depletion-aggregation process of SXC, it was hypothesized that the flow rate depends on the surface area of ​​one membrane layer. The flow rate of 7 mL min, previously determined for the MA15 device, -1 The optimal flow rate is 1.426 mL min -1 ·cm -2 The aim was to apply the same surface area dependent flow rate to the MA100 device. Since the membrane surface area per layer is four times larger, the flow rate for the MA100 device is 1.426 mL·min -1 ·cm -2 The flow rate is 28 mL min -1 corresponds to (Table 1).

[0065] However, with the viscous PEG buffer, the -1 The pressure limit was reached when applying 1.426 mL min −1. The UV cell and fractionator of the chromatography system contribute to the pressure. To circumvent this technical limitation, the chromatography system was opened after the column position and fractionated manually. This adjustment allowed the MA100 device to deliver 1.426 mL min −1. -1 ·cm-2 (28mL min -1 ) was applied. SXC runs were performed using the MA15 and MA100 devices at a flow rate of 1.426 mL min -1 ·cm -2 1.73 x 10 7 TU·mL -1 No significant difference in infectious titer was detected when LV batches with titers of 1000 or 10000 were used (Figure 3B). With the MA100, an infectious LV recovery of 72.79 ± 12.92% was obtained. These results confirm the hypothesis that the flow rate must be scaled to the surface area of ​​the membrane layer. The flow rate through the stationary phase appears to be a crucial factor for the success of the purification. The same flow rate (mL min) was applied to the MA100 device. -1 ), the same feed is distributed over a larger surface area and thus a larger number of pores compared to MA15. Because the average pore size remains unchanged, the flow rate through the pores decreases and falls below the optimal flow rate through the stationary phase, achieving efficient capture of LVs during loading and release during elution.

[0066] For the SXC runs on MA100, internal and external mixing of the LV solution with PEG buffer was performed as previously done for the MA15 runs and as known in the art. Briefly, LV solution (1.64 × 10 7 TU·mL -1The LVs (titer of 10 ... Once aggregates form, the free energy of the system is already reduced, leading to less efficient depletion interactions between the LV and the membrane, resulting in losses in the flow-through. These observations highlight the highly dynamic nature of this chromatographic method, as already recognized by the importance of flow rate.

[0067] Next, 1.426 mL·min -1 ·cm -2We investigated protein and dsDNA removal rates using the MA15 and MA100 devices at the same surface-area-dependent flow rates (Figure 4). The dsDNA and protein concentrations in the loading and elution fractions are listed in Table 2. Overall, high protein removal rates (80.51 ± 2.22% for MA15 and 76.72 ± 6.81% for MA100) were observed. Silver-stained SDS-PAGE gels confirm the measurements, revealing a significant amount of protein contaminants in the loading material and the removal of most of the protein impurities in the flow-through (Figure 4B). The elution fraction showed protein bands for the lentiviral vector structural proteins and almost no contaminating proteins. The dsDNA removal rates were 55.44 ± 12.58% for MA15 and 62.91 ± 8.06% for MA100. These results demonstrate that comparable impurity removal rates can be achieved with both device types. The effective removal of impurities is driven by the significant size difference between LV and contaminating proteins and DNA.

[0068] [Table 2]

[0069] Next, various loading volumes ranging from 100 mL to 700 mL (equivalent to 50 mL to 350 mL of LV solution) were tested in the MA100 device. The LV batches were 1.35 × 10 7 TU·mL -1 and 1.14 × 10 10 VP·mL -1Previous SXC experiments using MA100 were performed with a loading volume of 200 mL. The flow-through and elution fractions of all runs were analyzed. No increase in the amount of LVs in the flow-through was observed with increasing loading volume (Figures 5A and 5D). These findings are supported by images of HEK293T cells that did not express GFP after transduction with the flow-through fraction (Figure 5B). In contrast, HEK293T cells transduced with LVs from the elution fraction showed GFP expression (Figure 5C). When a large amount of LVs was loaded (Figure 5D), elution of the captured LVs became nearly impossible, resulting in low recovery in the elution fraction.

[0070] Loading approximately 200 mL achieved the highest infectious recovery of approximately 60% and particulate recovery of approximately 100% (Figures 5E and 5F). 9 TU and 1.14 x 10 12 The loading amount of VP was defined. In contrast to conventional chromatography methods, SXC does not rely on a functionalized stationary phase (resulting in limited binding sites). As revealed by visualization of LV on the membrane, LVs are primarily captured on the upper membrane layer. Therefore, LV breakthrough was not observed. Therefore, the membrane capacity for SXC cannot be defined as a 10% LV breakthrough rate. Instead, various loading capacities and the success of LV elution are analyzed to determine the loading amount at which LV recovery in the elution is satisfactory. In previous experiments (Figure 3B), 4.10 × 10 VP were used for MA100 and MA15, respectively. 8 TU and 1.60 x 10 9Loading of TU revealed that approximately four times more LVs could be loaded into the MA100 device compared to the MA15 device. The amount of LV loaded was less than in previous studies using CAR-T-based LVs at higher LV titers in the loading material. These differences in upstream materials likely account for the discrepancies in results, and the ideal loading volume may need to be determined separately for each target. Another reason is that the LV distribution on the membrane may be uneven when using the standard housing of the MA100, as discussed in Example 3, resulting in inefficient elution of overloaded membrane areas.

[0071] To further analyze the presented approach of scaling flow rate according to the membrane surface area of ​​the first layer, scale-down experiments were performed using an axial PP15 device at three different flow rates, each with N=3, and scale-up experiments were performed using a radial 5-inch device at two different flow rates, each with N=1. Further runs at different flow rates were performed with MA15 and MA100 modules, each with N=3, to complement the data.

[0072] According to the literature, this is the first study using membrane capsules for SXC and the largest membrane module used for this method to date, with a loaded LV volume of 0.98 L. Pressure limitations have frequently been discussed as a potential obstacle to the scale-up of SXC. As previously reported, viscous buffers generate higher pressures compared to traditional chromatographic methods such as anion exchange chromatography, and pressure increases during loading are frequently reported. During two scale-up runs using 5-inch capsules, pressure increases were observed from 0.4 bar to 0.8 bar (run 1) and from 0.5 bar to 0.7 bar (run 2). Because the device's pressure limit is 4 bar, pressure was not a limiting factor during the scale-up runs under the test conditions.

[0073] Figure 9 shows infectious and particulate recoveries and impurity removal for four different device scales plotted against various surface area specific flow rates.

[0074] Plotting the recoveries of infectious and particle titers for the device scales tested against various surface area specific flow rates reveals that LV recovery decreases significantly when flow rates are reduced below a critical minimum flow rate (Figure 9A, Figure 9B). LV recovery appears to asymptotically approach a maximum value of 3.5 mL min -1 ·cm -2 Although LV recovery may decrease at flow rates above 1.4 mL min, there are technically feasible limits due to the maximum flow rate of the system and the maximum pressure of the module. Although further investigation is needed to confirm this observation, it is clear that too low a surface-area-dependent flow rate will significantly decrease LV recovery, and generally, for successful scale-up of SXC, a flow rate of approximately 1.4 mL min is required. -1 ·cm -2 A surface-area-dependent flow rate of 1.43 mL min is required. Scaling the flow rate and the LV volume loaded according to the membrane area of ​​the chromatography module keeps the processing time for a complete chromatography run constant, so that an SXC run with either an MA15 or 5-inch capsule requires a flow rate of 1.43 mL min -1 ·cm -2 This short processing time is particularly beneficial for fragile enveloped viruses and viral vectors, enabling a fast and efficient DSP process. By using a scaling approach with minimal surface-area-dependent flow rates, we were able to achieve reproducible LV recovery rates with the SXC for four different module sizes. The highest LV volume purified by the SXC was 980 mL, with a recovery rate of 68%. This corresponds to an overall scaling factor of 98 compared to the smallest device, the PP15 (Table 1). The dsDNA removal rate shows a decreasing trend with increasing surface-area-specific flow rates (Figure 9C). Achieving both high LV recovery and dsDNA removal rates requires a flow rate of 1.4 mL min -1 ·cm -2 to 2.5 mL min -1 ·cm -2A surface area-specific flow rate between 0.01 and 0.1 is favorable, reaching a dsDNA removal rate of approximately 51%. Protein removal was not affected by flow rate and was consistent across various module sizes, with a protein removal rate of approximately 84% (Figure 9D). Overall, good impurity removal rates were achieved, and the DSP process is likely to be followed by subsequent ultrafiltration and diafiltration steps to remove residual PEG and further enhance product purity.

[0075] Example 3: The effect of membrane housing design on SXC performance. After identifying the process parameters critical for the scale-up of SXC using axial chromatography devices, the effect of membrane housing design on the capture of LV on the membrane and the performance of SXC was investigated.

[0076] The membrane chromatography device used in this study is operated by axial flow through the membrane stack from above and has a low bed height (the height of the stacked membrane layers). In addition to the lower bed height, the incident flow area is larger than that of a resin column. To avoid drift and efficiently use the entire membrane area, uniform flow distribution over the entire membrane area is required. This is achieved by a distributor structure inside the lid that spreads the fluid over the membrane and a collector structure inside the table that collects the fluid. The housing shape has a significant impact on the transport of fluid through the membrane, so housing design should play an important role in the development of a chromatography process.

[0077] Lentiviral vectors were visualized using MA100 housing to evaluate the distribution of LVs on the membrane before elution. SXC was administered at 1.426 mL min -1 ·cm -2 A volume of 160 mL was loaded, which corresponds to a flow rate of 7.27 x 10 9 VP·mL -1 and 2.71 x 10 7 TU·mL -1This corresponds to a volume of 80 mL of LV solution with an infectious titer of 10 ...

[0078] In the second SXC run, the device configuration was inverted, and the table (Figure 1H) was used as the lid, and vice versa. Thus, the incoming fluid was distributed by the radial and circular distribution channels. Figure 6B shows that LVs were more uniformly distributed on the membrane layer. The presence of LVs on the first membrane layer was visible, which is comparable to the findings when using MA15 (Figure 2). The modified lid and table configuration in this run significantly improved fluid distribution on the membrane. These findings demonstrate that a lid with radial and circular distribution channels is better suited to spreading fluid on the membrane than a roughened structure with thick bridges (Figure 1G). Dark spots within bright regions indicate the presence of air bubbles, which prevented fluid from reaching the membrane in these areas. Air bubbles reduce the utilization of the surface area they occupy, resulting in reduced recovery of target species. A larger pressure drop across the membrane could remove the bubbles.

[0079] We constructed a prototype with radial and circular distribution channels in the lid and table. This housing configuration also resulted in uniform distribution of LV on the membrane (Figure 6C). Some air bubbles (dark spots) were present within the device. Comparing Figures 6B and 6C, the LV appears to be better distributed in the prototype housing. This is likely because the distribution channel design not only distributes fluid evenly to the top of the membrane stack, but also more efficiently collects fluid from the membrane and directs it toward the outlet of the table. This LV visualization experiment demonstrates that the design of the membrane module is important to achieve uniform distribution of fluid on the membrane so that the entire membrane area can be utilized.

[0080] LVs were purified by SXC using the three device configurations described above. In this experiment, the MA100 membrane, lid, and table were assembled into a stainless steel holder (Figure 1E). Therefore, comparability with the overmolded MA100 device used in previous experiments is limited. LV recovery rates for various MA100 housing configurations are shown in Figure 7. All previous experiments used the standard configuration, which served as a comparison with the inverted and prototype configurations. Virus solutions purified with the standard and inverted configurations (Figure 7A) yielded 1.02 × 10 10 VP·mL -1 and a total particle titer of 2.39 x 10 7 TU·mL -1 The virus solution purified with the prototype device and standard configuration (Figure 7B) had an infectious titer of 3.03 × 10 7 TU·mL -1 at a concentration of 4.32 x 10 9 VP·mL -1 The total particle titer was 1.0001.

[0081] The inverted configuration yielded higher infectious LV recovery and total LV particle recovery compared to the standard configuration, but the differences were not significant. The LV recovery rates of the prototype configuration were also not significantly different from those of the standard configuration, although the standard deviation was lower for the prototype configuration. These findings indicate that the use of inlet and outlet distribution structures can generally produce more consistent LV recovery rates. The prototype and inverted configurations reduce the inlet dead volume, reduce backmixing effects, and promote narrower residence time distributions. Considering the high LV recovery rates, uniform LV distribution on the membrane, and lowest dead volume, the prototype device with flow distributor plates in the lid and table is a favorable configuration compared to the axial MA100 device.

[0082] Example 4: Steric exclusion chromatography of adeno-associated virus (AAV) AAV8 lysates were thawed on ice and conditioned with 1x SLPB (SXC lysate purification buffer). For clear filtration, 5g / L (dry weight) of Sartoclear Dynamics Lab filter aid was added to the lysate and gently mixed. Vacuum filtration was performed using a 0.22µM PES (polyethersulfone) membrane bottle-top filter (Sartorius AG). Lysates were not stored at 4°C for longer than 1 day. During the run, AAV8 samples were kept on ice, while A1 and B1 were set to room temperature.

[0083] SXC experiments were performed using an AEKTA chromatography system at 12 mL / min, which corresponds to a flow rate of 2.44 L min -1 ·cm -2 This surface area specific flow rate is within the range of the determined optimal flow rate and corresponds to a surface area specific flow rate of 1.0 L min -1 ·cm -2 The critical minimum flow rate was exceeded (Figure 9). To reduce the pressure, fractions were manually collected through the outlet valve without flowing through the entire system. Maximum pressure: 0.65 MPa (system pressure).

[0084] A proprietary Octet R8 label-free protein BLI analysis system was used to quantify the AAV viral capsid concentration in the samples. AAV8 standards from Progen (AAV8 empty capsid, 6.6 × 10) were used. 12 vp / mL, Catalog No. 66V080).

[0085] [Table 3]

[0086] 15% PEG 4k in-line SXC purification of AAV8 by AEKTA via Hydrosart 10242 (MA15). The respective AAV recovery and SXC chromatograms are shown in Figures 11 and 12, respectively.

[0087] Flow rate: 12mL / min.

[0088] Equilibration: Buffer A1 and Buffer B1 (44.1%) were mixed to reach a final PEG concentration of 15% PEG 4000 (44.1% Buffer B1 was mixed with 55.9% Buffer A) and the membrane was equilibrated with 50 CV (20.5 mL).

[0089] Sample application / flow-through Buffer A2 and Buffer B1 (44.1%) were mixed to reach a final PEG concentration of 15% PEG 4000 (44.1% Buffer B1 mixed with 55.9% Buffer A), and the membrane was loaded with 121.95 CV (50 mL, 27.95 mL AAV sample + 22.05 mL 34% PEG 4k stock).

[0090] Column cleaning Buffer A1 and Buffer B1 (44.1%) were mixed to reach a final PEG concentration of 15% PEG 4000 (44.1% Buffer B1 was mixed with 55.9% Buffer A), and the membrane was washed with 39.59 CV (15 mL).

[0091] Elution Isocratic elution was performed with 100% A1 (PEG-free buffer) applying 48.78 CV (20 mL, 10 x 2 mL fractions) in the upflow direction.

[0092] 10% in-line SXC purification of AAV8 by AEKTA via Hydrosart 10242 (MA15). The respective AAV recovery and SXC chromatograms are shown in Figures 13 and 14, respectively.

[0093] Flow rate: 12mL / min.

[0094] Equilibration Buffer A1 and Buffer B1 (29.41%) were mixed to reach a final PEG concentration of 10% PEG 4000 (29.41% Buffer B1 was mixed with 70.59% Buffer A) and the membrane was equilibrated with 50 CV (20.5 mL).

[0095] Sample application / flow-through Buffer A2 and Buffer B1 (29.41%) were mixed to reach a final PEG concentration of 10% PEG 4000 (29.41% Buffer B1 mixed with 70.59% Buffer A), and the membrane was loaded with 121.95 CV (50 mL, 35.29 mL AAV sample + 14.71 mL 34% PEG 4000 stock).

[0096] Column cleaning Buffer A1 and Buffer B1 (29.41%) were mixed to reach a final PEG concentration of 10% PEG 4000 (29.41% Buffer B1 was mixed with 70.59% Buffer A), and the membrane was washed with 39.59 CV (15 mL).

[0097] Elution Isocratic elution was performed with 100% A1 (PEG-free buffer) applying 48.78 CV (20 mL, 10 x 2 mL fractions) in the upflow direction.

[0098] From this series of experiments, it can be concluded that 15% PEG 4000 performs better and results in higher AAV recovery compared to 10% PEG 4000.

[0099] Example 5: Visualization of AAVs Objective: Visualization of AAV on SXC membrane using Alexa 647 dye → preliminary observation for further SXC optimization.

[0100] Materials and Devices Capture select Alexa Fluor 647 anti-AAVX conjugate (product number: 7233522100) from Thermo Scientific.

[0101] Two Hydrosart 10242 membranes (5 layers) were assembled in an MA15 (CV 0.41 mL) housing with a stainless steel holder.

[0102] Maximum pressure: 0.65MPa (system pressure).

[0103] Preparation of AAV8 samples 120 mL of AAV8 lysate was thawed (on ice) and mixed with 13.3 mL of 10x SLPB (→ 1x SLPB conditioning).

[0104] Alexa Fluorophore 647 antibody (44.4 μL to a 1:3000 dilution) was added to the AAV lysate, mixed and incubated overnight at 4° C. in the dark.

[0105] The next day, the AAV was clarified using diatomaceous earth and a 0.22 μm bottle-top vacuum filter (Sartoclear Dynamics Lab, Sartorius).

[0106] SXC was performed as described above with a final PEG 4000 concentration of 15%. AAV was visualized before (after column wash) and after elution (Figure 10). The fluorescence of each SXC membrane layer was analyzed using a BioRad ChemiDoc MP imaging system. As observed with LV, AAV was also found primarily on the first one to two membrane layers.

[0107] Conclusion: Steric exclusion chromatography has demonstrated potential as a gentle purification method for large enveloped viral vectors. However, scale-up has not yet been explored, raising the question of how to address this challenge. Visualization of LVs and AAVs on the membrane revealed that SXC is a surface-directed process, i.e., LVs and AAVs are primarily captured on the upper membrane layer. The present invention demonstrated that flow rates must be scaled along with the membrane area of ​​the first layer. Scale-down and scale-up experiments demonstrated that a certain critical minimum surface-area-dependent flow rate is required to achieve reproducible LV recovery for four different device scales tested, with an overall scale-up factor of 98. In the largest-scale run, 980 mL of LVs was successfully purified using a radial device geometry, and further scale-up could be achieved by using larger capsule or cassette modules. Various loading volumes were investigated, but no LV breakthrough was observed with increasing volume. However, LV elution from the overloaded membrane region was nearly impossible, indicating the optimal amount of LV to be loaded. By modifying the housing design of the MA100 module, the flow distribution was improved, resulting in a uniform distribution of LVs on the membrane. Using the improved housing prototype, overloading of the membrane area was more likely to be avoided, offering the possibility of loading more LVs. Overall, the scalability of SXC using membrane modules has been demonstrated in this invention, which lays the foundation for the future industrial applicability of this method.

Claims

1. 1. A method for purifying viruses, virus-like particles, extracellular vesicles, proteins having a diameter of 5 nm or greater, protein complexes having a diameter of 5 nm or greater, nucleic acids having a diameter of 5 nm or greater, protein-nucleic acid complexes having a diameter of 5 nm or greater, and / or molecules having a diameter of 5 nm or greater by filtration-assisted polyalkylene glycol precipitation, comprising: (a) providing a filtration module; wherein the filtration module is (i-α) a stack or two separate stacks of 2 to 9 layers of stabilized nonionic hydrophilic porous structures, said layers being placed on top of each other, wherein each of said stack(s) exhibits a homogeneous flow resistance and has a total thickness of 300 μm or more and 2000 μm or less, or (i-β) one monolithic stabilized non-ionic hydrophilic porous structure or two separate monolithic stabilized non-ionic hydrophilic porous structures, each of said porous structure(s) exhibiting a uniform flow resistance and having a total thickness of 300 μm or more and 2000 μm or less; (ii) a housing containing the stack(s) of porous structure layers or the monolithic porous structure(s), the housing comprising a fluid inlet and a fluid outlet, wherein fluid entering the filtration module through the fluid inlet must pass through the stack(s) of porous structure layers or the monolithic porous structure(s) and exit the filtration module through the fluid outlet, and the housing is adapted to uniformly distribute a feed solution entering the housing via the fluid inlet over the entire accessible surface area of ​​a first layer of the stack(s) of porous structure layers or over the entire accessible surface area of ​​the monolithic porous structure(s); equipped with; (b) introducing a feed solution comprising the virus, virus-like particle, extracellular vesicle, protein having a diameter of 5 nm or more, protein complex having a diameter of 5 nm or more, nucleic acid having a diameter of 5 nm or more, protein-nucleic acid complex having a diameter of 5 nm or more, and / or molecule having a diameter of 5 nm or more through the fluid inlet into a first layer of the stack(s) of layers of the porous structure or the monolithic porous structure(s) over an accessible surface area of ​​1 cm ２ At least 1.0 mL per minute (1.0 mL min －１ ・cm －２ ) into the filtration module at a surface area dependent flow rate, thereby attaching the viruses, virus-like particles, extracellular vesicles, proteins having a diameter of 5 nm or greater, protein complexes having a diameter of 5 nm or greater, nucleic acids having a diameter of 5 nm or greater, protein-nucleic acid complexes having a diameter of 5 nm or greater, and / or molecules having a diameter of 5 nm or greater to the porous structure; (c) eluting the viruses, virus-like particles, extracellular vesicles, proteins having a diameter of 5 nm or more, protein complexes having a diameter of 5 nm or more, nucleic acids having a diameter of 5 nm or more, protein-nucleic acid complexes having a diameter of 5 nm or more, and / or molecules having a diameter of 5 nm or more from the porous structure, thereby recovering purified viruses, virus-like particles, extracellular vesicles, proteins having a diameter of 5 nm or more, protein complexes having a diameter of 5 nm or more, nucleic acids having a diameter of 5 nm or more, protein-nucleic acid complexes having a diameter of 5 nm or more, and / or molecules having a diameter of 5 nm or more; A method comprising:

2. The method according to claim 1, wherein the laminate of porous structure layers is a laminate of 3 to 5 porous structure layers.

3. The method according to claim 1 or 2, wherein the porous structure has a pore size of 1 μm to 5 μm.

4. The method according to any one of claims 1 to 3, wherein each of the porous structure layers has a thickness of 100 µm to 300 µm.

5. The top layer of the stack of porous structure layers has an accessible diameter of more than 25 mm and a thickness of 4.91 cm ２ 5. The method of claim 1, wherein the surface of the substrate has an accessible surface area of ​​greater than 1000 nm.

6. The method according to any one of claims 1 to 5, wherein the porous structure is a stabilized cellulose membrane.

7. The method according to any one of claims 1 to 6, wherein the filtration module is an axial flow module, a capsule module, or a cassette module.

8. 8. The method of any one of claims 1 to 7, wherein prior to step (b), the feeding solution and the buffer are mixed in a buffer:feed solution ratio of 4:1 to 1:

4.

9. 9. The method of claim 8, wherein the buffer comprises a polyalkylene glycol having 2 or 3 carbon atoms in the repeating unit.

10. 10. The method of claim 9, wherein the polyalkylene glycol is polyethylene glycol (PEG).

11. 11. The method of claim 10, wherein the PEG is a PEG having a molar mass between 400 g / mol and 12000 g / mol.

12. 12. The method of claim 10 or 11, wherein the buffer comprises the polyalkylene glycol or PEG in an amount such that the final polyalkylene glycol or PEG concentration after mixing with the feed solution is between 4% and 20% by weight.

13. The method according to any one of claims 1 to 12, wherein the elution in step (c) is carried out in an upflow direction.

14. 14. The method according to any one of claims 1 to 13, wherein the virus or respective virus-like particle is selected from the group consisting of lentivirus, baculovirus, orf virus, adeno-associated virus (AAV), influenza virus (such as influenza A virus), yellow fever virus, human papillomavirus, vaccinia virus, adenovirus, hepatitis virus, poliovirus, rabies virus, rotavirus, rubella virus, Zika virus, and respective virus-like particles.

15. 1. A filtration module comprising: (a-α) a stack or two separate stacks of 2 to 9 layers of stabilized nonionic hydrophilic porous structures, said layers being placed on top of each other, wherein each of said stack(s) exhibits a homogeneous flow resistance and has a total thickness of 300 μm or more and 2000 μm or less, or (a-β) one monolithic stabilized non-ionic hydrophilic porous structure or two separate monolithic stabilized non-ionic hydrophilic porous structures, each of said porous structure(s) exhibiting a uniform flow resistance and having a total thickness of 300 μm or more and 2000 μm or less; (b) a housing containing the stack(s) of porous structure layers, the housing comprising a fluid inlet and a fluid outlet, wherein fluid entering the filtration module through the fluid inlet must pass through the stack(s) of porous structure layers and exit the filtration module through the fluid outlet, and the housing is adapted to uniformly distribute a feed solution entering the housing via the fluid inlet over the entire accessible surface area of ​​a first layer of the stack(s) of porous structure layers; A filtration module comprising:

Images