WO2021188629A1

WO2021188629A1 - Label-free high-throughput isolation of large cho cell clusters

Info

Publication number: WO2021188629A1
Application number: PCT/US2021/022703
Authority: WO
Inventors: Taehong Kwon; Jongyoon Han
Original assignee: Massachusetts Institute Of Technology
Priority date: 2020-03-18
Filing date: 2021-03-17
Publication date: 2021-09-23

Abstract

The invention provides a method of separating single CHO cells from CHO cell clusters, said method comprising supplying a suspension/fluid comprising single CHO cells and CHO clusters into a microfluidic device comprising a curvilinear microchannel having a trapezoidal cross section. In an embodiment, the device has one inlet and two outlets and a trapezoidal cross-section having an inner depth of 239 μm, an outer depth of 83 μm and a width of 1000 μm.

Description

LABEL-FREE HIGH-THROUGHPUT ISOLATION OF LARGE CHO CELL CLUSTERS

Field of the Invention

The present invention relates to a method of separating Chinese Hamster Ovary (CHO) cell clusters from single CHO cells.

Background

Background to the Invention

In recent decades, Chinese Hamster Ovary (CHO) cells have been the most prevalent mammalian host cells for producing biotherapeutics. CHO cells are generally in suspension in bioreactors that can range from small- to large-scale. Since CHO cells are anchorage- dependent cells in origin, large CHO cell clusters are often formed during bioreactor cultivation. This can be problematic because cells at the centre of clusters may not be exposed to sufficient nutrients and oxygen. As a result, cell growth and protein production may be significantly reduced.

Various approaches have been used to reduce these cell clusters in bioreactors: anti aggregation agents (e.g., dextran sulfate, heparin, and suramin), cell dissociation enzymes (e.g., trypsin), membrane filters, and cell engineering. Although anti-aggregation agents and cell dissociation enzyme can be used to dissociate clusters, these culture additives may affect product quality and volumetric productivity. Additionally, removal performance may vary depending on the specific cell line and culture conditions. Membrane filtration approaches such as cell strainers may suffer from filter fouling/clogging and reduced recovery efficiency. More importantly, there is no filtration method to remove cell clusters and simultaneously maintain cells during continuous perfusion operation, since existing hollow fibre membrane filters keep both cell clusters and single cells in perfusion bioreactors. As another approach, the CHO cell line may be engineered to reduce tendency of forming clusters by regulating biological pathways. However, this development takes a large amount of time and effort, and engineered cell line could still produce aggregates, depending on culture conditions.

There is therefore a need to easily separate CHO cell clusters from CHO cells without requiring chemical additives, filters, or cell line engineering. In addition, precise and efficient manipulation of high-density particle suspensions is critical for applications in many fields. Although membrane-based filtration is commonly used for separation and manipulation of high-density particle suspensions, it has many limitations, such as membrane fouling and clogging. These challenges become more severe for higher density suspensions such as blood and CHO cell culture in bioprocessing, highlighting the need for the development of membrane-less particle manipulation techniques with simplicity, good cell viability, and high throughput.

Inertial microfluidics has been used to control particle motion in Newtonian fluids at high throughput. Recently, the combination of this fluid inertia and fluid elasticity has been utilized to manipulate particles further in microchannels (i.e., elasto-inertial microfluidics). This has led applications in particle focusing and separation. However, in most prior experiments, viscoelasticity of the cell solution was artificially induced by adding polymeric additives, which is highly undesirable in cell suspensions. However, some cell suspensions when at high cell density exhibit intrinsic viscoelasticity ( e.g ., blood and culture suspensions), implying that direct elasto-inertial manipulation may be possible. However, the importance of cell density in elasto-inertial particle focusing is not fully understood and has not been fully investigated. Existing studies in cell focusing and separation in elasto-inertial flows have been limited to relatively low cell densities (< 3% packed cell volume (PCV)) (e.g., Holzner, G., Stavrakis, S. & DeMello, A. Elasto-inertial Focusing of Mammalian Cells and Bacteria Using Low Molecular, Low Viscosity PEO Solutions. Anal. Chem. 89, 11653-11663 (2017)).

Accordingly, there is a need for a method to separate CHO cells from CHO cell clusters, particularly at high cell suspension densities.

Theoretical Background: Elasto-inertial focusing

The combination of inertial and elastic focusing at finite fluid inertia and elasticity regimes enables unique rigid particle or cell focusing at high throughput (Elasto-inertial focusing). Two non-dimensional numbers, the channel Reynolds number (Re) and Weissenberg number (1/1//), measure the inertial and elastic effect of the fluid, respectively:

where Pf is the fluid density; v is the average fluid velocity;

is the hydraulic diameter of the channel; is the zero-shear rate viscosity of the fluid.

where ^λ is the relaxation time of the fluid; is the characteristic shear rate.

The elasto-inertial effect is often expressed using another non-dimensional number,

Elasticity number , as follows:

Viscoelasticity, one of the unique properties of non-Newtonian fluid, is a key element of elasto-inertial focusing. It causes elastic lift force on particles to drive them laterally to the equilibrium positions. Elastic lift force arising from this viscoelasticity of the fluid is

expressed as follows:

where is the elastic lift coefficient; a is the spherical diameter of the particle; is the

polymeric contribution to the zero-shear rate solution viscosity

. The elastic lift force is applied in the direction toward the region with a smaller shear rate, typically the centre of the microfluidic channel.

In addition, the fluid with finite inertia in microchannels causes inertial lift forces on the cells to migrate them laterally toward several equilibrium positions along the channel. This inertial migration arises from two dominant size-dependent hydrodynamic forces: shear-gradient lift force and wall-induced lift force. They result from parabolic flow profile of pressure-driven flows and particle-wall interaction, respectively. The net lift force, ^FL, is expressed as

where is the lift coefficient of depending on channel Reynolds number and

particle lateral position in the channel.

In addition to elastic and lift forces, curved microchannels such as spirals induce lateral secondary-flow (Dean flow) drag force by generating two counter-rotating vortexes in the cross-section so that cell focusing along the channel can be further controlled laterally. This drag force is expressed as:

where is

the average transverse Dean velocity; is the non-dimensional number representing the strength of the Dean flow.

Summary of the Invention

The present inventors have surprisingly found that a microfluidic device comprising a curvilinear microchannel having a trapezoidal cross section can effectively separate CHO cells from CHO cell clusters under a wide range of conditions, including at high cell suspension density. This provides for high-throughput isolation of large CHO cell clusters from bioreactors while retaining single cells.

The separation at high and ultra-high cell densities is possible due to the microfluidic device providing elasto-inertial particle focusing at ultra-high cell density ( e.g 29.7 PCV%), without requiring any buffer additives. This may also be used to advantageously reduce the cell density of an ultra-high-density solution.

The invention therefore provides the below numbered clauses.

1. A method of separating Chinese Hamster Ovary (CHO) cell clusters from single CHO cells comprising the step of: supplying a suspension/fluid comprising single CHO cells and CHO clusters into an inlet of a microfluidic device, where the microfluidic device comprises: i. at least one inlet; ii. a curvilinear microchannel fluidly connected to the at least one inlet, the curvilinear microchannel having a trapezoidal cross section defined by a radial inner side, a radial outer side, a bottom side and a top side, where the radial inner side and radial outer side have different heights; and iii. a first outlet located on the radial outer side of the curvilinear microchannel; and iv. a second outlet located on the radial inner side of the curvilinear microchannel, wherein the suspension/fluid is supplied to the microfluidic device at a flow rate that isolates particles along portions of the cross-section of the microchannel based on particle size, where a first set of particles flow along the radial outer side of the microchannel to a first outlet, and a second set of particles flow along the radial inner side of the microchannel to a second outlet.

2. The method of Clause 1 , wherein the suspension/fluid supplied to the microfluidic device has a packed cell volume of 0.2% or greater, optionally 1% or greater, more optionally 5% or greater, further optionally 10% or greater, further optionally still 20% or greater.

3. The method of Clause 2, wherein the suspension/fluid supplied to the microfluidic device has a packed cell volume of 20% or greater.

4. The method of any one of the preceding clauses, wherein the suspension/fluid supplied to the microfluidic device has a packed cell volume of from 20% to 40%, optionally from 20% to 35%, such as from 20% to 30%.

5. The method of any one of the preceding clauses, wherein the microfluidic device comprises a third outlet at a middle portion of the curvilinear microchannel, and where a third set of particles flow along the middle portion of the curvilinear microchannel.

6. The method of any one of the preceding clauses, wherein the suspension/fluid flow through the curvilinear microchannel immediately before the first, second, and when present third, outlets comprises a concentration of CHO cells at the centre of the curvilinear microchannel, by width, that is greater than the concentration of CHO cells adjacent to the radial inner and radial outer wall of the curvilinear microchannel.

7. The method of any one of the preceding clauses, wherein the suspension/fluid supplied to the microfluidic device comprises from 0.05 wt% to 2 wt. % polyethylene oxide.

8. The method of any one of the preceding clauses, wherein the suspension/fluid flowing through the first, second, and when present the third outlet, each have a different CHO cell density to that of the suspension/fluid supplied to the microfluidic device.

9. The method of any one of the preceding clauses, wherein the CHO cells present in the suspension/fluid have a diameter in the range of 16 to 22 pm, optionally in the range of 16 to 20 pm, more optionally in the range of 16 to 18 pm, such as about 17 pm.

10. The method of any one of the preceding clauses, wherein the Reynolds number (Re) of the flow through the microfluidic device is from 20 to 330, optionally from 120 to 330, more optionally from 120 to 230. 11. The method of any one of the preceding clauses, wherein the first set of particles have an average particle size larger than the second set of particles.

12. The method of Clause 1 or 11 , wherein the first set of particles comprises CHO cell clusters, optionally wherein the CHO cell clusters have a diameter of greater than 30 pm.

13. The method of any one of Clauses 1 , 11 or 12, wherein the second set of particles comprises single CHO cells.

14. The method of any one of Clauses 1 and 11 to 13, wherein the second set of particles is substantially free of CHO cell clusters having a diameter of greater than 30 pm, optionally wherein less than 10 wt%, such as less than 5 wt%, less than 3 wt%, less than 2 wt%, less than 1 wt% or less than 0.1 wt% of the mass of CHO cells in the second set of particles is composed of CHO cell clusters having a diameter of greater than 30 pm.

15. The method of any one of the preceding clauses, wherein the height of the radial inner side of the curvilinear microchannel is greater than the height of the radial outer side of the curvilinear microchannel.

16. The method of Clause 15, wherein the ratio of the height of the radial inner side of the curvilinear microchannel to the height of the radial outer side of the curvilinear microchannel is greater than 2:1 , such as from about 2.5:1 to about 3:1.

17. The method of Clause 15, wherein the ratio of the height of the radial inner side of the curvilinear microchannel to the height of the radial outer side of the curvilinear microchannel is from about 1.1 :1 to about 3:1 , optionally from about 1.1 :1 to about 2.9:1 , such as about 1 .2:1 to about 2.1 :1.

18. The method of any one of Clauses 1 to 14, wherein the height of the radial outer side of the curvilinear microchannel is greater than the height of the radial inner side of the curvilinear microchannel.

19. The method of Clause 18, wherein the ratio of the height of the radial outer side of the curvilinear microchannel to the height of the radial inner side of the curvilinear microchannel is greater than 2:1 , such as from about 2.5:1 to about 3:1. 20. The method of Clause 18, wherein the ratio of the height of the radial outer side of the curvilinear microchannel to the height of the radial inner side of the curvilinear microchannel is from about 1.1 :1 to about 1 .7:1 , such as about 1.3:1 to about 1.5:1.

21. The method of any one of the preceding clauses, wherein the ratio W:H is from about 2:1 to about 10:1 , such as about 3:1 to about 5:1 ; where H is the greater of the height of the radial inner side of the curvilinear microchannel and the height of the radial outer side of the curvilinear microchannel; and W is the width of the curvilinear microchannel

22. The method of Clause 21 , wherein the ratio W:H is from about 4:1 to about 9:1 .

23. The method of any one of the preceding clauses, wherein the width of the curvilinear microchannel is substantially uniform along its length.

24. The method of any one of the preceding clauses, wherein:

(a) the height of the radial inner side of the curvilinear microchannel is from about 20 to about 300 microns;

(b) the height of the radial outer side of the curvilinear microchannel is from about 20 to about 300 microns; and/or

(c) the width of the microchannel in the curvilinear microchannel is from about 100 to about 2000 microns.

25. The method of Clause 24, wherein:

(a) the height of the radial inner side of the curvilinear microchannel is from about 200 to about 280 microns, such as from about 220 to about 260 microns;

(b) the height of the radial outer side of the curvilinear microchannel is from about 50 to about 120 microns, such as about 70 to about 90 microns; and/or

(c) the width of the microchannel in the curvilinear microchannel is from about 800 to about 1200 microns.

26. The method of Clause 24, wherein:

(a) the height of the radial inner side of the curvilinear microchannel is from about 60 to about 100 microns, such as from about 70 to about 90 microns;

(b) the height of the radial outer side of the curvilinear microchannel is from about 110 to about 150 microns, such as about 120 to about 140 microns; and/or (c) the width of the microchannel in the curvilinear microchannel is from about 400 to about 800 microns, such as about 500 to about 700 microns.

27. The method of Clause 24, wherein:

(a) the height of the radial inner side of the curvilinear microchannel is from about 70 to about 250 microns, for example about 160 to about 250 microns, such as from about 175 to about 240 microns;

(b) the height of the radial outer side of the curvilinear microchannel is from about 70 to about 160 microns, for example about 90 to about 160 microns, such as about 100 to about 150 microns; and/or

(c) the width of the microchannel in the curvilinear microchannel is from about 500 to about 1800 microns, for example about 800 to about 1800 microns, such as about 900 to about 1600 microns.

28. The method of any one of the preceding clauses, wherein the curvilinear microchannel has a cross-sectional area of from about 0.5 x 10⁶ to about 3 x 10⁶ pm², such as about 1.5 x 10⁶ to about 3 x 10⁶ pm².

29. The method of any one of the preceding clauses, wherein the curvilinear microchannel is a spiral microchannel, optionally wherein the spiral microchannel comprises at least 2 loops, such as at least 4 loops, for example at least 7 loops.

30. The method of any one of the preceding clauses, wherein the curvilinear microchannel has a radius of curvature of from about 2.5 mm to about 25 mm.

31. The method of any one of the preceding clauses, wherein the top and/or bottom side of the curvilinear microchannel has a linear cross section.

32. The method of any one of the preceding clauses, wherein the top and/or bottom side of the curvilinear microchannel has either (a) a concave cross section, or (b) a convex cross section.

33. The method of any one of the preceding clauses, wherein the flow rate of the suspension/fluid through the curvilinear microchannel is from about 1 to about 30 mL/min, such as about 5 to about 15 mL/min, for example about 10 mL/min. 34. The method of any one of the preceding clauses, wherein the flow rate of the suspension/fluid through the curvilinear microchannel is from about 0.5 to about 10 mL/min, optionally from about 3 to about 10 mL/min, more optionally from about 4 to about 8 mL/min.

35. The method of any one of the preceding clauses, wherein the suspension/fluid is supplied to the inlet of the microfluidic device by a pump configured to pump the suspension/fluid through the microfluidic device.

36. The method of any one of the preceding clauses, which is performed in a bioreactor, where the suspension/fluid comprising single CHO cells and CHO clusters is taken from the bioreactor, optionally wherein the output from the first outlet is removed from the bioreactor and/or the output from the second outlet is re-supplied into the bioreactor.

37. The method of any one of the preceding clauses, wherein the method comprises supplying the suspension/fluid to two or more microfluidic devices arranged in series, where an outlet of a first microfluidic device is connected to an inlet of a second microfluidic device.

38. The method of Clause 34, wherein the height of the radial inner side of the curvilinear microchannel of the first microfluidic device is greater than the height of the radial outer side of the curvilinear microchannel of the first microfluidic device, and the height of the radial inner side of the curvilinear microchannel of the second microfluidic device is lower than the height of the radial outer side of the curvilinear microchannel of the second microfluidic device.

39. The method of Clause 34 or 35, wherein the flow rate through the first microfluidic device is greater than the flow rate through the second microfluidic device, optionally wherein the flow rate through the first and second microfluidic devices are both as defined in Clause 33 or 34.

40. The method of any one of the preceding clauses, wherein:

(i) the flow rate of the suspension/fluid through the curvilinear microchannel is from about 0.5 to about 10 mL/min; and

(ii) the suspension/fluid supplied to the microfluidic device has a packed cell volume of from 20% to 40%.

The invention provides a number of advantages. Reducing the density of an ultra-high-density CHO cell suspension using the method of the invention is highly advantageous for a number of reasons. In particular, it provides cost- effective selective separation of large cell clusters without using physical barriers (membrane filters) or other external forces (e.g., acoustic or electrical forces). The cell separation is enabled by a mass-producible plastic device and pressure-driven flow. In addition, the method provides continuous density reduction without dilution. This is particularly useful in certain industrial processes/applications such as cell retention for perfusion bioreactors, which requires processing of intact cell culture suspensions from bioreactors.

Moreover, the method of the invention can also provide focusing of a cell suspension into a specific part of the cross section of a curvilinear microchannel contained within the microfluidic device. This can be particularly beneficial for the following reasons.

1) It allows selection of the desired output from the microfluidic device. Particular cell populations near the inner or outer wall can be chosen as the desired output stream. For example, the suspension near the outer wall is likely to contain more CHO cell clusters. Depending on the application, the inner or outer output could be preferred depending on whether it is desirable to remove more clusters or more of the smaller cell population.

2) The method can provide three distinct cell populations at once: 1. High density, 2. Reduced density with smaller average particle size (e.g. near the inner wall), 3. Reduced density with larger average particle size (e.g. near the outer wall)

3) Cell manipulation or separation is possible at very high cell densities (>30 PCV%) as the centreline focusing arising from high viscoelasticity of high-density cell suspensions is maintained due to dominant elastic lift force generated in spiral channels.

Brief Description of the Figures

Figures 1A-D show the removal of CHO cell clusters using spiral inertial microfluidic chip. (A) shows the input of a CHO cell suspension containing clusters and single cells into the inlet of the spiral inertial microfluidic chip. The CHO cell clusters were collected at an outer outlet, while the single cells were collected at an inner outlet. (B) shows the cross-section of the trapezoidal microchannel, and shows the location of the CHO cell clusters and single cells. (C) shows a photograph of the chip with one inlet and two outlets. (D) shows an image of a sample from the outer outlet of the chip containing CHO cell clusters.

Figures 2A-D show the assessment of cluster removal performance of the microfluidic device. (A) shows the steps of an automated analysis of the area of the CHO cell clusters using ImageJ as described in Example 1. (B) shows histograms for the size of clusters obtained from the inlet and outer outlet, based on image analysis. (C) shows the cluster removal efficiency and enrichment factor. (D) shows the biocompatibility in terms of cell growth, metabolism, and IgG production, as determined by performing batch culture of control (non-processed) and processed CHO cells (i.e., cells that had been passed through the microfluidic device).

Figures 3A-B show the focusing of high-density ( >10 PCV%) CHO cell suspensions in the microfluidic device. (A) shows the cell focusing in the microfluidic device at different cell densities. The images were captured from the last two loops of the microfluidic device of Example 2. (B) shows the contribution of the elastic lift force (F_ei_) arising from viscoelasticity of high-density CHO cell suspensions to cell focusing in the channel. Inertial lift (Fi_) and Dean drag (F_D) forces are not shown.

Figures 4A-D show cell focusing for high-density culture suspensions and PEO-added culture suspensions in the microfluidic device as described in Example 2. (A) shows the cell focusing shift at different densities (PCV%) in high-density culture suspension. The images were captured from the last two loops of the microfluidic device. (B) shows the lateral cell focusing profile from the top view for high-density culture suspensions. (C) shows the cell focusing shift for different PEO concentrations. The CHO cell density was 0.5 million cells/mL (<0.1 PCV%). The images were captured from the last two loops of the microfluidic device. (D) shows the lateral cell focusing profile from the top view for PEO-added culture suspension.

Figure 5 shows the centreline focusing of CHO cells at 35.1 PCV% cell density.

Figures 6A-C show cell focusing at different flow rate and cross-sectional dimension conditions. (A) shows cell focusing shift at three different input flow rates (2, 6, and 10 mL/min) in a 179 pm/110 pm/1500 pm (inner depth/outer depth/width) microfluidic device. The high-density (PCV 26.1%) culture suspension was used. The images were captured from the last two loops of the microfluidic device. (B) shows the lateral cell focusing profile from the top view according to different input flow rates. (C) shows the position of the focusing peak at different PCV% in six microfluidic devices described in Example 2.

Figure 7 shows the cell focusing images at different cell densities for six microfluidic devices as described in Example 2.

Figure 8 shows centreline focusing of an ultra-high density (30.7 PCV%) CHO cell suspension using the microfluidic device of Example 1.

Figure 9 shows density reduction based on fluid split and input cell density. IO and 00 refer to the inner outlet and outer outlet, respectively. (A) shows a computational flow simulation of two different fluid split cases (IO flow%:00 flow%=18%:82% and 82%:18%). (B) shows the actual density reduction at different fluid split values. Input cell culture at 17.1 PCV% was flowed into the device at 10 mL/min. Error bars, standard deviation (n = 3). (C) shows the actual density reduction at three different input cell densities (17.9, 22.9, and 27.3 PCV%). The input flow rate was 10 mL/min, and fluid split was 11 %:89% (I0%:00%). Error bars, standard deviation ( n = 3).

Figure 10 shows cluster removal at high cell density (27.3 PCV%). (A) Images for outlet samples collected from the microfluidic device. (B) Comparison of the number of large cell clusters (>10 cells) per 45 sample images between the inner and outer outlets. Error bars, standard deviation (n = 3).

Figure 11 shows how the packed cell volume (solid volume fraction) of a suspension against CHO cell concentration for different cell sizes.

Figure 12 shows a three-outlet configuration for a microfluidic device that is useful for centreline cell focusing.

Figure 13 shows density reduction using a cascaded (series) configuration as in Example 3. (A) shows a schematic of two microfluidic devices in series to reduce cell density of ultra- high cell density suspensions. (B) shows a photo of two spiral devices in cascaded (series) configuration. (C) shows the density of samples collected from the input and outlets of the two microfluidic devices in series. Error bars, standard deviation ( n = 3). Figure 14 shows CHO cell focusing in the microfluidic device arranged second in a series configuration as in Example 3. (A) shows the microfluidic device. (B) shows the cell focusing behaviour at three different input flow rates. The input cell density was 16 PCV%. The image was captured from the last four loops of the microfluidic device.

Detailed Description of the Invention

In embodiments herein, the word “comprising” may be interpreted as requiring the features mentioned, but not limiting the presence of other features. Alternatively, the word “comprising” may also relate to the situation where only the components/features listed are intended to be present (e.g., the word “comprising” may be replaced by the phrases “consists of” or “consists essentially of”). It is explicitly contemplated that both the broader and narrower interpretations can be applied to all aspects and embodiments of the present invention. In other words, the word “comprising” and synonyms thereof may be replaced by the phrase “consisting of” or the phrase “consists essentially of” or synonyms thereof and vice versa.

In embodiments herein, various features may be described in the singular or the plural. It is herein explicitly contemplated that references to the singular are to be understood as including the plural, and references to the plural are to be understood as including the singular, unless such an interpretation would be technically illogical.

The invention provides a method of separating Chinese Hamster Ovary (CHO) cell clusters from single CHO cells comprising the step of: supplying a suspension/fluid comprising single CHO cells and CHO clusters into an inlet of a microfluidic device, where the microfluidic device comprises: i. at least one inlet; ii. a curvilinear microchannel fluidly connected to the at least one inlet, the curvilinear microchannel having a trapezoidal cross section defined by a radial inner side, a radial outer side, a bottom side and a top side, where the radial inner side and radial outer side have different heights; and iii. a first outlet located on the radial outer side of the curvilinear microchannel; and iv. a second outlet located on the radial inner side of the curvilinear microchannel, wherein the suspension/fluid is supplied to the microfluidic device at a flow rate that isolates particles along portions of the cross-section of the microchannel based on particle size, where a first set of particles flow along the radial outer side of the microchannel to a first outlet, and a second set of particles flow along the radial inner side of the microchannel to a second outlet.

As well as separating CHO cells from CHO cell clusters, the microfluidic device may also focus CHO cells towards a certain part of the cross section of the microfluidic device. This focusing may be density-dependent, i.e. dependent on the density of CHO cells in the suspension/fluid passing through the microfluidic device. In general, when the CHO cells are focused towards the outer wall or near the centreline of the microfluidic device, the fluid stream closer to the inner wall comprised a lower amount of CHO cell clusters. As such, separation of CHO cells from CHO cell clusters, as used herein, also refers to focusing of a CHO cell stream to a particular part of the cross section of the microfluidic device. This cell focusing is typically centreline focusing, i.e. focusing CHO cells towards the centre of the cross section of the microfluidic device, but may also be outer wall focusing. By centreline focusing, it is meant that the suspension/fluid flowing through the curvilinear microchannel (such as near the end of the curvilinear microchannel, e.g., immediately before the outlets) comprises a concentration of CHO cells at the centre of the curvilinear microchannel, by width, that is greater than the concentration of CHO cells adjacent to the radial inner and radial outer wall of the curvilinear microchannel. In other words, centreline focusing refers to there being more CHO cells present in the middle of the curvilinear microchannel (by width) than at either wall. This results in a “band”, as visible in the images in Figure 7. As is clear from the results in Figure 10, cell focusing may provide a substantial cell density reduction and cluster removal using ultra-high cell density suspensions as the input fluid.

The microfluidic device used in the method of the invention comprises a curvilinear microchannel, e.g., a microchannel having a spiral shape that may have multiple loops. For example, the microfluidic device may comprise at least 2 loops, such as at least 3 loops, at least 4 loops, at least 5 loops, at least 6 loops, for example at least 7 loops.

The microfluidic device comprises at least one inlet and at least two outlets that are connected to the curvilinear microchannel such that fluid may pass through the inlet, along the curvilinear (e.g., spiral) microchannel and to the at least two outlets. The at least two outlets comprise a first outlet and a second outlet. The first outlet is located on the radial outer side of the curvilinear microchannel, and the second outlet is located on the radial inner side of the curvilinear microchannel. In this way, particles that are not dispersed evenly across the cross section of the curvilinear microchannel may be collected in different amounts/proportions at each of the outlets. As such, in some embodiments of the invention the suspension/fluid flowing through each outlet may have a different CHO cell density to that of the suspension/fluid supplied to the microfluidic device.

The curvilinear microchannel has a trapezoidal cross section. The combination of the curved (e.g., spiral) nature of the curvilinear microchannel and the trapezoidal cross section act to separate particles of different sizes flowing through the microchannel, and at certain cell densities, to focus particles of a similar size into specific regions of the cross section of the curvilinear microchannel. This allows for separation of particles based on size, such as separation of CHO cell clusters from single CHO cells. It also allows focusing of CHO cells at higher cell densities.

Thus, when a suspension/fluid comprising CHO cells and CHO cell clusters is supplied to the microfluidic device at a suitable flow rate (e.g., 0.5 to 30 mL/min) a first set of particles may flow along the radial outer side of the microchannel to a first outlet, and a second set of particles may flow along the radial inner side of the microchannel to a second outlet. In some embodiments, the device may additionally comprise a third outlet, that may be located at a middle portion of the curvilinear microchannel (Figure 12). This may be useful when a suspension/fluid is supplied under conditions that result in a third set of particles flowing along the middle portion of the curvilinear microchannel (e.g., centreline focusing at high CHO cell densities).

In some embodiments that may be mentioned herein, the suspension/fluid supplied to the microfluidic device has a packed cell volume of 0.2% or greater (e.g., 1% or greater, 5% or greater, 10% or greater, 15% or greater, 16% or greater, 17% or greater, 18% or greater 19% or greater, such as 20% or greater). The maximum packed cell volume of the suspension/fluid supplied to the microfluidic device may be 40% (e.g., 35%, 30%, or 25%).

For the avoidance of doubt, it is herein explicitly contemplated that any end point of a range may be combined with any other end point of a range defining the same variable. For example, the suspension/fluid supplied to the microfluidic device may have a packed cell volume of: from 0.2% to 1%, from 0.2% to 5%, from 0.2% to 10%, from 0.2% to 15%, from 0.2% to 16%, from 0.2% to 17%, from 0.2% to 18%, from 0.2% to 19%, from 0.2% to 20%, from 0.2% to 25%, from 0.2% to 30%, from 0.2% to 35%, from 0.2% to 40%, from 1% to 5%, from 1% to 10%, from 1% to 15%, from 1% to 16%, from 1% to 17%, from 1% to 18%, from 1% to 19%, from 1% to 20%, from 1% to 25%, from 1% to 30%, from 1% to 35%, from 1% to 40%, from 5% to 10%, from 5% to 15%, from 5% to 16%, from 5% to 17%, from 5% to 18%, from 5% to 19%, from 5% to 20%, from 5% to 25%, from 5% to 30%, from 5% to 35%, from 5% to 40%, from 10% to 15%, from 10% to 16%, from 10% to 17%, from 10% to 18%, from 10% to 19%, from 10% to 20%, from 10% to 25%, from 10% to 30%, from 10% to 35%, from 10% to 40%, from 15% to 16%, from 15% to 17%, from 15% to 18%, from 15% to 19%, from 15% to 20%, from 15% to 25%, from 15% to 30%, from 15% to 35%, from 15% to 40%, from 16% to 17%, from 16% to 18%, from 16% to 19%, from 16% to 20%, from 16% to 25%, from 16% to 30%, from 16% to 35%, from 16% to 40%, from 17% to 18%, from 17% to 19%, from 17% to 20%, from 17% to 25%, from 17% to 30%, from 17% to 35%, from 17% to 40%, from 18% to 19%, from 18% to 20%, from 18% to 25%, from 18% to 30%, from 18% to 35%, from 18% to 40%, from 19% to 20%, from 19% to 25%, from 19% to 30%, from 19% to 35%, from 19% to 40%, from 20% to 25%, from 20% to 30%, from 20% to 35%, from 20% to 40%, from 25% to 30%, from 25% to 35%, from 25% to 40%, from 30% to 35%, from 30% to 40%, and from 35% to 40%.

As described herein, centreline focusing may occur primarily for viscoelastic fluids. Ultra-high cell density suspensions may inherently exhibit viscoelastic properties. Viscoelasticity may also be induced by addition of polyethylene oxide (PEO). Thus, in some embodiments of the invention, the suspension/fluid supplied to the microfluidic device comprises from 0.05 wt% to 2 wt. % polyethylene oxide.

As will be appreciated by a person skilled in the art, different CHO cell lines may have different cell diameters. The method of the invention may in general be performed with any CHO cell line, and may particularly be performed with CHO cells having a diameter in the range of 14 to 22 pm (e.g., in the range of 16 to 22 pm, 16 to 20 pm, in the range of 16 to 18 pm, such as about 17 pm). In some embodiments that may be mentioned herein, the Reynolds number (Re) of the flow through the microfluidic device is from 20 to 330, for example 120 to 330, such as 120 to 230.

As discussed herein, in some embodiments of the invention, the first set of particles may have an average particle size larger than the second set of particles. In other words, the number/proportion of CHO cell clusters running along the outer wall may be greater than the number/proportion of CHO cell clusters running along the inner wall, such that the average particle size of particles running along the outer wall is greater than those running along the inner wall. Thus, first set of particles may comprise CHO cell clusters, which may have a diameter of greater than 30 pm. The second set of particles may comprise single CHO cells. In some embodiments of the invention that may be mentioned herein, the second set of particles may be substantially free of CHO cell clusters having a diameter of greater than 30 pm. In such embodiments, less than 10 wt%, such as less than 5 wt%, less than 3 wt%, less than 2 wt%, less than 1 wt% or less than 0.1 wt% of the mass of CHO cells in the second set of particles may be composed of CHO cell clusters having a diameter of greater than 30 pm.

The curvilinear microchannel has a trapezoidal cross section. In some embodiments of the invention that may be mentioned herein, the height of the radial inner side of the curvilinear microchannel may be greater than the height of the radial outer side of the curvilinear microchannel. In some embodiments of the invention that may be mentioned herein, the ratio of the height of the radial inner side of the curvilinear microchannel to the height of the radial outer side of the curvilinear microchannel may be greater than 2:1 , such as from about 2.5:1 to about 3:1. In some embodiments of the invention that may be mentioned herein, the ratio of the height of the radial inner side of the curvilinear microchannel to the height of the radial outer side of the curvilinear microchannel may be from about 1.1 :1 to about 2.9:1 , for example from about 1.1 :1 to about 2.5:1 , such as about 1.2:1 to about 2.1 :1.

In some embodiments of the invention that may be mentioned herein, the height of the radial outer side of the curvilinear microchannel may be greater than the height of the radial inner side of the curvilinear microchannel. In some embodiments of the invention that may be mentioned herein, the ratio of the height of the radial outer side of the curvilinear microchannel to the height of the radial inner side of the curvilinear microchannel may be greater than 2:1 , such as from about 2.5:1 to about 3:1. In some embodiments of the invention that may be mentioned herein, the ratio of the height of the radial outer side of the curvilinear microchannel to the height of the radial inner side of the curvilinear microchannel may be from about 1.1 :1 to about 1.7:1 , such as about 1.3:1 to about 1.5:1.

Typically, the curvilinear microchannel will have a width greater than the height of the inner or outer wall. Thus, in some embodiments of the invention that may be mentioned herein, the ratio W:H may be from about 2:1 to about 10:1 , such as about 3:1 to about 5:1 ; where H is the greater of the height of the radial inner side of the curvilinear microchannel and the height of the radial outer side of the curvilinear microchannel; and W is the width of the curvilinear microchannel.

In some embodiments, the ratio W:H may be from about 4:1 to about 9:1.

The width of the curvilinear microchannel may be substantially uniform along its length.

Exemplary geometries for the curvilinear microchannel are provided below.

1. The height of the radial inner side of the curvilinear microchannel may be from about 20 to about 300 microns; the height of the radial outer side of the curvilinear microchannel may be from about 20 to about 300 microns; and/or the width of the microchannel in the curvilinear microchannel may be from about 100 to about 2000 microns.

2. The height of the radial inner side of the curvilinear microchannel may be from about 200 to about 280 microns, such as from about 220 to about 260 microns; the height of the radial outer side of the curvilinear microchannel may be from about 50 to about 120 microns, such as about 70 to about 90 microns; and/or the width of the microchannel in the curvilinear microchannel may be from about 800 to about 1200 microns.

3. The height of the radial inner side of the curvilinear microchannel may be from about 60 to about 100 microns, such as from about 70 to about 90 microns; the height of the radial outer side of the curvilinear microchannel may be from about 110 to about 150 microns, such as about 120 to about 140 microns; and/or the width of the microchannel in the curvilinear microchannel may be from about 400 to about 800 microns, such as about 500 to about 700 microns. 4. The height of the radial inner side of the curvilinear microchannel may be from about 70 to about 250 microns, for example about 160 to about 250 microns, such as from about 175 to about 240 microns; the height of the radial outer side of the curvilinear microchannel may be from about 70 to about 160 microns, for example about 90 to about 160 microns, such as about 100 to about 150 microns; and/or the width of the microchannel in the curvilinear microchannel may be from about 500 to about 1800 microns, for example about 90 to about 160 microns, such as about 900 to about 1600 microns.

In some embodiments of the invention, the curvilinear microchannel may have a cross- sectional area of from about 0.5 x 10⁶ to about 3 x 10⁶ pm², such as about 1 .5 x 10⁶ to about 3 x 10⁶ pm².

The curvilinear microchannel may have a radius of curvature of from about 2.5 mm to about 25 mm. In some embodiments, the top and/or bottom side of the curvilinear microchannel may have a linear cross section. In some embodiments, the top and/or bottom side of the curvilinear microchannel may have either (a) a concave cross section, or (b) a convex cross section.

As mentioned herein, the suspension/fluid comprising CHO cells and CHO cell clusters is supplied to the microfluidic device at a suitable flow rate (e.g., 0.5 to 30 mL/min) for a first set of particles to flow along the radial outer side of the microchannel to a first outlet, and for a second set of particles to flow along the radial inner side of the microchannel to a second outlet. In some embodiments of the invention, the flow rate of the suspension/fluid through the curvilinear microchannel is from about 1 to about 30 mL/min, such as about 5 to about 15 mL/min, for example about 10 mL/min. In some embodiments of the invention, the flow rate of the suspension/fluid through the curvilinear microchannel is from about 0.5 to about 10 mL/min, optionally from about 3 to about 10 mL/min, more optionally from about 4 to about 8 mL/min. The endpoints of these ranges may be combined in the manner explained above in relation to cell density ranges, i.e. any of these end points may be combined with any other of the end points. The suspension/fluid may be supplied to the inlet of the microfluidic device by a pump configured to pump the suspension/fluid through the microfluidic device, which may also control the flow rate.

As will be understood by a person skilled in the art, the method of the invention may be performed in a bioreactor, where the suspension/fluid comprising single CHO cells and CHO clusters is taken from the bioreactor (e.g., from the reaction medium). In this case, the output from the first outlet may be removed from the bioreactor and/or the output from the second outlet may be re-supplied into the bioreactor (e.g., into the reaction medium). Alternatively, it may be desirable to return the output from the first outlet to the bioreactor, and/or remove the output from the second outlet to the bioreactor.

The separation efficiency of the method of the invention may be improved by using two (or more) microfluidic devices as described herein in a cascaded (i.e. series) configuration. In this case, an outlet of a first microfluidic device is fluidly connected to an inlet of a second microfluidic device. In a particular embodiment of the invention that may be mentioned herein, the height of the radial inner side of the curvilinear microchannel of the first microfluidic device is greater than the height of the radial outer side of the curvilinear microchannel of the first microfluidic device, and the height of the radial inner side of the curvilinear microchannel of the second microfluidic device is lower than the height of the radial outer side of the curvilinear microchannel of the second microfluidic device.

In another particular embodiment of the invention, when the method involves a series configuration of two (or more) microfluidic devices in series, the flow rate through the first microfluidic device may be greater than the flow rate through the second microfluidic device.

Applications in which microfluidic CHO cell cluster removal is useful

First, cluster removal may be applied to cell passaging to prolong the cultivation with improved cell growth, viability, and protein production. Cell lines are maintained by passaging existing highly confluent cell culture into fresh medium at low cell concentration before cells become nonviable. Since the old cell culture could contain cell clusters, those clusters could be removed by the spiral chip before cell passaging. Second, cluster removal may be performed intermittently during bioreactor cultivation to enhance cell growth, viability, and protein production. The CHO cell clusters could be formed during cultivation and can be selectively removed by the spiral chip during cultivation.

Third, the cluster removal may be performed continuously during perfusion culture. Since the microfluidic device can retain the single cells through the one of the outlets and remove clusters through another outlet, it can be used as a cell retention device that also removes clusters.

Fourth, the technology can be combined with small dead cell removal by the spiral microfluidic device for effective removal of dead cells produced during bioreactor cultivation. As dead cells could be either small due to apoptosis or trapped in the cell clusters, this combined approach could achieve effective removal of dead cells produced during bioreactor cultivation. Removal of dead cells may improve product quality or productivity during biomanufacturing.

Fifth, the cluster removal can be applied to other cell types, such as circulating tumour cell clusters or large stem cells / tissue. For example, large circulating tumour cell clusters could be separated from small blood cells (liquid biopsy).

Sixth, the method may provide a population of single cells with improved purity at one of the outlets of the device, which is useful for single cell research.

Finally, the cluster removal may be performed as a pre-filter to improve accuracy of cell analysing equipment, such as flow cytometer.

When the method of the invention is used to reduce the cell density of an ultra high-density cell suspension, it has the following additional advantages.

First, the method can be used for the retention of cells in a high-cell-density perfusion bioreactor. In such cases, most CFIO cells can be continuously recycled back to the original bioreactor while a cell-limited stream can be collected separately. This enables a high cell density to be maintained in perfusion bioreactors. The technology can be also applied to other cell types, for example FIEK293T cells for virus particle production (gene therapy), stem cells for cell therapy, and microalgae for biomass feedstock production. Second, the reduction of cell density may be helpful in technologies that have an inherent cell concentration (density) capacity for effective and efficient operation. This may be especially useful in cell manipulation and downstream culture analysis. For example, the method of the invention may be used to reduce cell density before a suspension is passed through another microfluidic device that is not able to handle such high cell densities.

The below Examples illustrate the invention and are not to be construed as limitative.

Examples

General Methods

CHO cell culture and culture sample preparation

CHO-S cells (Freestyle™ CHO-S Cells, R80007, Thermo Fisher Scientific, USA) were grown in glass spinner flasks (4500-500, Corning, USA) in a 5% C0₂ incubator. The cells were seeded at 0.3 million cells/mL to a 250-310 ml. culture medium (Freestyle™ CFIO Expression Medium, 12651022, Thermo Fisher Scientific, USA) and maintained for up to 5-7 days. The culture parameters such as density, viability, nutrients/metabolites concentration, aggregate formation, pH, and ions were regularly monitored using an automated cell culture analyzer (FLEX2, NovaBiomedical, USA). In order to prepare high-density cell cultures, the cell culture was centrifuged (200 g for 3 minutes), and cell-free solution was removed. Packed cell volume (PCV%) was calculated as below to confirm high cell density (>30 million cells/mL) accurately in the presence of cell clusters. 100

The prepared high-density culture suspension was transferred into either 5mL macrotubes (for large sample volume of >1 mL; 470225-006, VWR, USA) or hematocrit tubes (for small sample volume of <1 mL; 10007500-C/5, Drummond, USA). Subsequently, the tubes containing culture suspension were centrifuged at 1800 g for 5 minutes to settle cells.

PEO sample preparation

To test the focusing of CFIO cells in polymer-based viscoelastic fluids, polyethylene oxide (PEO) was used (372781 -5G, viscosity average molecular weight of 1 ,000 kDa, MilliporeSigma, USA). PEO-added cell culture solution was prepared at two different wt% (0.1 wt% and 1wt%) concentrations in culture medium (Freestyle™ CHO Expression Medium, Thermo Fisher Scientific, USA). Based on an earlier study (Flolzner, G., Stavrakis, S. & DeMello, A. Elasto-lnertial Focusing of Mammalian Cells and Bacteria Using Low Molecular,

Low Viscosity PEO Solutions. Anal. Chem. 89, 11653-11663 (2017)), the relaxation time (' ) of the PEO-added (1wt%) culture was estimated to be 1.58 ms. The original cell culture without PEO at 0.5 million cells/mL was centrifuged for 3 minutes at 200 g, and cell-free solution was then removed. Subsequently, the PEO containing cell-free medium was transferred to make viscoelastic cell culture suspension at 0.5 million cells/mL.

Microfluidic device design and fabrication

The moulds for the microfluidic devices were designed using a 3D modeller (Rhinoceros, Robert McNeel & Associates, USA). The dimensions of other spiral devices are described below (Table 1). The moulds were fabricated with aluminium (Whits Technologies, Singapore). In order to make microfluidic devices, standard soft lithography using polydimethylsiloxane (PDMS, Sylgard 184, Dow Corning, USA) was used. The PDMS was poured onto the mold and then cured at 150 °C for 20 minutes. Subsequently, the micropattern ed PDMS piece was removed from the mould, and input and output reservoirs were made using punches (Integra LifeSciences, USA). It was then bonded to the PDMS- coated (500 pm thick) glass substrate using oxygen plasma treatment (FEMTO SCIENCE, Korea). The prepared microfluidic device was placed on a hotplate at 95 °C overnight for complete bonding.

For a cascaded (series) configuration, two different spiral devices were used in series as described in Example 3. The inner outlet of the first device was connected to the input of the second device. The culture sample collected from the inner outlet of the second device was regarded as a final output.

Observation of cell focusing and characterization of microfluidic device performance In order to deliver the culture suspension to the device, silicone tubing (EW-96410-14, Cole- Parmer, USA) were inserted to each input and output reservoirs of the device. The culture suspension was then loaded to a syringe (Becton Dickinson, USA), and a syringe pump (PHD 2000, Harvard Apparatus, USA) delivered the culture suspension to the device at various flow rates (up to 10 mL/min). The focusing behaviour of the cells was monitored under the inverted microscope (1X51 , Olympus, Japan) using a high-speed imaging camera (Phantom v9.1 , Vision Research, USA). Fluid simulation was performed using COMSOL Multiphysics 5.5 (COMSOL, Inc., USA). To characterize the density reduction quantitatively, the packed cell volume of the input and output samples was measured after centrifuging the cells in either macrotubes or capillary tubes. To vary the fluid split (inner outlet flow% vs. outer outlet flow%), the fluidic resistance of the outlet was modulated using a smaller- diameter tubing (EW-06420-02, Cole-Parmer, USA). By attaching this tubing and changing its length, the fluid split was controlled. The volume of the cell suspension collected from each outlet was measured to obtain the actual fluid split values. The cell reduction efficiency of the microfluidic device was obtained using the following equation:

where the output cell density corresponds to the cell density of the outlet of the device which contains reduced cell density. Typically, this outlet is the inner outlet of the spiral device when the PCV% is high ( e.g ., >20 PCV%). This reduction efficiency is often used as cell retention efficiency (%) in the context of cell retention for perfusion culture.

Example 1 : Separation of CHO cell clusters and single CHO cells

Figures 1A-D show a spiral inertial microfluidic device for the removal of CHO cell clusters from culture. The device used in this Example had one inlet and two outlets and a trapezoidal cross-section (inner depth: 239 pm, outer depth: 83 pm, width: 1000 pm).

The cell culture containing CHO cell clusters was flowed into the inlet of the device by either a syringe pump or a peristaltic pump, and separation of single cells and clusters occurs along the channel due to size-dependent hydrodynamic forces. The CHO cell clusters occupy their equilibrium positions near the outer wall of the channel whereas the single cells are focused near the inner wall of the channel. The separated population (single cells and clusters) are collected at the inner and outer outlets, respectively. The image for the outer outlet sample in Figure 1 D was captured by an automated cell culture analyser (CDV, NovaBiomedical, USA), and clearly shows that the device can remove CHO cell clusters selectively.

This demonstrates that separation of CHO cell clusters and single cells requires only flow through the microfluidic device, and does not involve any membrane (and therefore avoids fouling/clogging and replacement of a membrane). Moreover, the separation can be applied to any CHO suspension culture and culture modes (e.g., batch or continuous perfusion cultures).

Cluster removal performance of the spiral chip was assessed by analysing images for the sampled cell culture (Figure 2A). The automated cell culture analyzer produces 40 images for one cell culture sample (left image). The boundary of the objects in the image was automatically detected using ImageJ (centre image). The area of the object can be calculated from the detected boundary (right image). Figure 2B shows a histogram of the cell culture samples obtained from the input and outer outlet of the chip. The input cell culture contained 2.3% of >600 pm² clusters whereas the outer outlet culture contained 30.1%, showing 13.2-fold enrichment. As shown in Figure 2C, the device demonstrated 47.6% cluster removal efficiency on average (n = 3) for clusters of 600-900 pm² and 50.5% removal efficiency for clusters of >900 pm². The average enrichment factors (n = 3) for 600- 900 pm² and >900 pm² clusters were 11.8 and 12.4, respectively. Figure 2D confirms that the processing of the cell culture by the spiral chip does not affect the cell growth, viability, metabolism, and IgG production. The cells were passaged after passing through the microfluidic device and cultured for a week. There was no significant difference in terms of growth, viability, metabolism, and IgG production between the control (non-processed) and processed groups.

The single device demonstrated high volumetric throughput (10 mL/min), high removal efficiency (50.5%), and biocompatibility. As the cluster removal is achieved due to size- dependent hydrodynamic forces arising from channel dimension and flow speed, the technology does not require any external forces (e.g., electrical or acoustic field) and physical barrier (e.g., membrane filter) to remove cell clusters. This label-free high- throughput isolation method could be used to remove large CHO cell clusters from small to large-scale bioreactors while maintaining single cells in any mode of operation (discontinuous/continuous) during bioprocessing. In addition, considering that the cell clusters are more likely to contain dead cells and small cell debris (<5 pm) are randomly dispersed in the microfluidic device without being focused, the technology can be used to remove dead cells and cell debris from cell culture. Comparative Example 1 : Existing cluster removal methods and their drawbacks

In contrast to these methods, the label-free high-throughput isolation method of the invention does not require any culture additives and does not have any physical membrane. The microfluidic device can be fabricated in an economical way (mass-producible using hard plastic materials) and is re-usable. It can be used from small to large-scale bioreactors in any mode of operation (discontinuous/continuous) during bioprocessing. In addition, the microfluidic device may be made by an additive manufacturing process, such as 3D printing. This provides a quick, cheap and reliable way to manufacture the microfluidic device with a high degree of accuracy.

Example 2: Focusing behaviour of CHO cells in hiqh-densitv and PEO-added cultures

An 8-loop spiral microfluidic device with trapezoidal cross-section (inner depth: 179 pm, outer depth: 110 pm, width: 1500 pm) was used to examine the focusing behaviour of CHO cells at three different cell densities (10.1 , 18.9, and 26.1 (packed cell volume % (PCV%)) (Figure 3A). The device had one inlet and two outlets. The concentrated CHO cells (average size: 17 pm) in culture medium at the desired density were flowed into the inlet of the device at 6 mL/min (Reynolds Number (Re) = 136). The focusing behaviour of the CHO cells was observed near the outlets (7^th and 8^th loops) (Figure 4A). The shift of focusing position was observed at all three different PCV%, creating density gradient laterally. At the lowest PCV% (10.1 PCV%), CHO cells occupied the space near the inner wall while making cell-free region near the outer wall. By contrast, at 18.9 PCV%, the CHO cells were found to be more focused near the outer wall, although the cells filled the entire channel in the lateral direction (no cell-free region), which was previously reported by Goh et al. (Goh, S., Tan, S. M., Tan, D. S. & Yang, Y. S. New Phenomenon : Outer Wall Focusing At High Cell Densities Enables High Performance in 20th International Conference on Miniaturized Systems for Chemistry and Life Sciences 270-271 (2016) and WO 2018/017022 A1). The peak intensity position near the outer wall was observed (Figure 4B). As the density was further increased to 26.1 PCV%, more focusing of the cells near the centreline of the channel was observed. Although cell-free region was still not formed, the cell density near both sides was lower than that near the centreline, producing a lateral cell density gradient.

To understand the association between unique cell focusing in the microfluidic device at different cell densities and viscoelasticity of the high-density culture, the inventors observed the focusing behaviour of the CHO cells in a typical viscoelastic fluid in the same microfluidic device. Poly-ethylene oxide (PEO) was added to the culture medium at different concentrations (0.1 wt% and 1 wt%) to make the fluid viscoelastic. To minimize the effect of interaction between cells, the input cell density of 0.5 million cells/mL (<0.1 PCV%) was used. At the same flow rate conditions (6 mL/min input flow rate), the focusing shift of the CHO cells was clearly observed (Figure 4C). The CHO cells in the culture medium without PEO were focused near the inner wall of the channel. By contrast, the cells in the medium containing 0.1 wt% PEO were shifted to near the outer wall of the channel. Interestingly, higher PEO concentration (1 wt%) caused the cells to be focused near the centreline of the channel, exhibiting the typical focusing behaviour of the cells in the highly viscoelastic fluid. Figure 4D shows the distinct cell focusing behaviour in different culture suspensions. Similar position shift was observed in culture suspension with different cell densities (Figure 4B).

Centreline focusing was demonstrated for cell suspensions having a PCV as high as 35.1% (Figure 5). The image was taken from the seventh loop of the microfluidic device using an input flow rate of 10 mL/min. Cell focusing study at varying Input flow rates and channel geometries

The effect of different flow rates and channel cross-sectional dimensions on focusing of high- density CHO cells was investigated. First, three different flow rates (2, 6, and 10 mL/min corresponding to Re = 45, 136, and 227, respectively) were tested using the high-cell- density (26.1 PCV%) culture suspension (Figure 6A). At this high cell density, all three flow rates formed the focused cell band away from the outer wall of the channel, focusing the cells near the centreline. As the input flow rate increased, the peak of the focused cell band moved toward the centreline of the channel (peak position of 392 pm, 608 pm, and 794 pm at 2, 6, and 10 mL/min, respectively) (Figure 6B), suggesting that more elastic lift force was generated due to increased shear rate

^{: "}). At 6 and 10 mL/min, reduced cell density near both wall sides of the channel was clearly observed.

Table 1 : microfluidic devices with different cross-sectional dimensions and flow conditions used for Figure 6C.

To examine whether the shifting of focused cell band is specific to a given microchannel, varying cross-sectional dimensions were tested: inner depth of 179-237 pm, outer depth of 110-148 pm, and width of 1000-1500 pm (Table 1). Regardless of the dimensions, the shift of the focusing position was observed with increased culture cell density (10.5% to 27.2%) (Figures 6C and 7). The shifting of the focused cell band from the inner wall to the outer wall, finally toward the centreline of the channel, was observed across the Re range of 126- 182. It is also noted that the centreline focusing of the CHO cells in the 1000 pm channels (Devices C, D, E, and F) was not dependent on the slope of the cross-section.

The microfluidic device of Example 1 was also used to provide cell density reduction. The device has the dimensions: inner wall: 239 pm, outer wall: 83 pm, and width: 1000 pm. It was confirmed that this device can be also used to reduce cell density from ultra-high- density CHO cell culture based on high-speed imaging for flow rates of both 2 mL/min and 6 mL/min (Figure 8). The input CHO cell density was 30.7 PCV% = 119.3 million cells/mL based on 17 pm CHO cells = 111.6 million cells/mL based on the density calibration using an automated cell culture analyser.

Density reduction study based on the fluid split at the bifurcation and input cell density

To examine the degree of density gradient formation from cell focusing, the device performance in density reduction was quantitatively characterized (/.e., cell reduction efficiency; see General Methods section). The CHO cell suspension at desired PCV% flowed into the microchannel at 10 mL/min input flow rate, and the PCV% values of the samples collected from both inner and outer outlets were measured.

First, modulating flow split at the bifurcation controlled the streamline boundary between two output flows (inner outlet (IO) and outer outlet (OO) flows) and affected density reduction efficiency. The fluid simulation example of streamline boundary modulation is shown in Figure 9A. Lower inner outlet flow% pushes the streamline boundary between the two outlet flows closer to the inner wall of the channel. When input cell density was 17.1 PCV% (corresponding to 66.5 million cells/mL with 17 pm CHO cells), more cells were focused near the outer wall of the channel; less cells were near the inner wall. Therefore, lower inner outlet flow% improved cell reduction efficiency with less cell density of the inner outlet flow. The reduction efficiency at three different fluid splits was as follows: (1) 57.6% at 11%:89% (I0%:00%), (2) 45.8% at 24%:76%, (3) 3.8% at 87%:13% (Figure 9B). Increased density of the outer outlet flow from 17.1 to 27.7 PCV% at 87%:13% fluid split confirmed the focusing of the cells near the outer wall at this high density. It is noted that even at the lowest inner outlet flow% (/.e., 11 %:89% split), the inner outlet flow rate was maintained high at approximately 1.1 mL/min. Assuming 17 pm CHO cells, the single-pass spiral operation reduced 66.5 million cells/mL (17.1 PCV%) to 28.2 million cells/mL (7.3 PCV%). The input cell density also affected the reduction efficiency of the spiral device. The reduction of cell density was observed for all three input densities (17.9, 22.9, and 27.3 PCV%) at 10 mL/min input flow rate and 11%:89% (I0%:00%) fluid split: (1) 17.9% to 7.3%, (2) 22.9% to 10.7%, and (3) 27.3% to 12.3% (Figure 9C). The reduction efficiency was 59.2%, 53.3%, and 54.9%, respectively. Compared with 17.9 PCV%, the reduction efficiency was decreased at both two higher densities (22.9 PCV% and 27.3 PCV%). However, the reduction efficiency at the highest 27.3 PCV% was slightly higher by 1 .6% than that at 22.9 PCV%. This shows that the reduction efficiency does not decrease linearly with increased cell density and more cells were moved to the outer outlet flow at 27.3 PCV%. The cell density collected in the outer outlet (OO) was also measured. Even with fluid split of I0%:00%=11%:89%, all cell densities of the outer outlet were higher than those of the input (feed), showing lateral suspension fractionation due to cell focusing. The PCV% values were converted to million cells/mL, assuming 17 pm CHO cells. Single-pass spiral operation reduced 106.1 million cells/mL (27.3 PCV%) to 47.9 million cells/mL (12.3 PCV%).

Samples of the cell suspension flowing through the inner outlet and outer outlet were also analysed to determine cluster removal and cell density reduction. Results are shown in Figure 10. (A) shows images for outlet samples collected from the microfluidic device. They were obtained by the automated cell culture analyser, which produces 45 samples per sample. (B) shows a comparison of the number of large cell clusters (>10 cells) per 45 sample images between the inner and outer outlets. Error bars, standard deviation (n = 3). It is clear from the images and results that the outer outlet sample contains substantially more cell clusters and has a higher cell density.

Comparative Example 2: Comparison between trapezoidal cross section and rectangular cross section

Goh et al. (WO 2018/017022 A1) previously reported CHO cell focusing on the outer wall of a rectangular cross section microfluidic device using CHO cells having a diameter of 15 pm at a concentration of approximately 100 million cells/mL, leading to a PCV of approximately

17% (Figure 11).

Goh reported that trapezoidal cross sectioned microfluidic devices provided for lower filtration efficiency than rectangular cross sectioned devices. The current inventors have surprisingly found that the method disclosed herein is able to provide a filtration efficiency higher than reported in Goh, using trapezoidal cross sectioned microfluidic devices. The inventors have further shown that this improved filtration efficiency is also obtained for higher flow rates, which is advantageous for processing a volume of CHO cell suspension more rapidly.

A comparison between the results in WO 2018/017022 A1 and the present invention is provided in Table 2 below.

Table 2

Example 3: Density reduction for ultra-hiqh-densitv suspension culture using cascaded

(series) configuration

Density reduction efficiency at high-density suspension can be further improved by using a cascaded configuration. Figure 13A shows the schematic of this cascaded configuration using two microfluidic devices. The dimensions for the first and second devices are 179 prn/110 pm/1500 pm (first device) and 84 pm/119 pm/600 pm (second device), inner depth/outer depth/width respectively. The first device reduces the cell density for the second device. Overall reduction efficiency can be further enhanced after cell reduction from the second device. At high cell density ( e.g >17 PCV%), the majority of the focused cells can be collected in the outer outlet of the device with modulated fluid split (Figure 9). The rest of the cells are collected in the inner outlet, according to the reduction efficiency of the device. These cells can be flowed again into the second device for further density reduction. The actual two devices in series are shown in Figure 13B. The inner outlet of the first device was connected to the inlet of the second device through a silicone tubing. Considering fluid split from the first device, the microfluidic device with smaller cross-sectional dimensions was used as the second device for effective cell focusing at lower input flow rates (Figure 14). The input flow rates for the first and second devices were 10 mL/min and 1.5 mL/min, respectively. The fluid split values for the first and second devices were I0%:00%=15%:85% and 26%:74%, respectively. The flow rate of the inner outlet of the second device (final output) was 0.39 mL/min (= 562 mL/day).

With the flow conditions described above, the CHO cell suspension at 29.7 PCV% flowed into the cascaded devices. The final output (inner outlet of the second device) had the lowest density (8.3 PCV%), achieving 72.1% overall reduction efficiency (Figure 13C). Compared with the earlier result (54.9% efficiency at 27.3 PCV%), the reduction efficiency was increased by 17.2 percentage points even with the increased input cell density from 27.3 to 29.7 PCV%. Using the fluid split and measured PCV%, the input density of the second spiral device was estimated to be 17.6 PCV%. Therefore, the reduction efficiencies of the first and second devices were 40.7% and 52.8% at 29.7 PCV% and 17.6 PCV%, respectively. Assuming 17 pm CHO cells, the single-pass operation in the devices in series reduced 115.5 million cells/mL (29.7 PCV%) to 32.3 million cells/mL (8.3 PCV%). As the cell density reduction depends on fluid split (Figure 9B), the reduction efficiency at this high cell density could be further improved when the lower flow rate for the inner outlet of the second device is used.

Comparative Example 3: Comparison with other technologies

Table 3. Comparison of cell retention technologies for perfusion culture

Tangential F ow Filtration Alternating Tangential Flow (Filtration) Based on the results in this study

The beneficial effects demonstrated in the above Examples have many immediate applications, including cell retention from continuous perfusion bioreactors. Continuous solid- liquid separation is routinely performed by hollow fiber membrane-based cell retention devices during perfusion culture in continuous biomanufacturing; the cell retention device retains cells in the bioreactor and separates out cell-less liquid containing toxic metabolites and therapeutic proteins from the bioreactor.

The above results show the potential enhancing cell retention capacity by microfluidic devices in the elasto-inertial fluid regime. Compared with other membrane-less cell manipulation techniques such as acoustic wave separators, centrifuges, hydrocyclones, and gravity settlers, the microfluidic cell retention device based on elasto-inertial cell focusing has many benefits. With its enhanced cell density capacity demonstrated in this study, the membrane-less microfluidic device can perform cell retention with advantages including removal of small dead cells/debris and low capital/operational expenses, which could lead to efficient and reliable high-density perfusion culture at various scales.

Using mammalian cell culture suspension (CFIO cells), it has been demonstrated that ultra- high-density suspension cells provide inherent viscoelasticity of the fluid, which allows unique focusing behaviour of the cells at high cell density suspensions. Using this unique elasto-inertial cell focusing, efficient cell clarification by the spiral microfluidic devices was demonstrated, with significant density reduction (72%) at ultra-high-density CFIO suspension (29.7 PCV%, corresponding to 115.5 million cells/mL with a 17 pm cell size). This clearly shows that high-density cell suspensions can be effectively manipulated in microchannels at high throughput utilizing inherent viscoelasticity, enabling various practical and industrial applications of high-throughput microfluidics.

Claims

2. The method of Claim 1 , wherein the suspension/fluid supplied to the microfluidic device has a packed cell volume of 0.2% or greater, optionally 1% or greater, more optionally 5% or greater, further optionally 10% or greater, further optionally still 20% or greater.

3. The method of Claim 2, wherein the suspension/fluid supplied to the microfluidic device has a packed cell volume of 20% or greater, optionally wherein the suspension/fluid supplied to the microfluidic device has a packed cell volume of from 20% to 40%, more optionally from 20% to 35%, such as from 20% to 30%.

4. The method of any one of the preceding claims, wherein the microfluidic device comprises a third outlet at a middle portion of the curvilinear microchannel, and where a third set of particles flow along the middle portion of the curvilinear microchannel.

5. The method of any one of the preceding claims, wherein the suspension/fluid flow through the curvilinear microchannel immediately before the first, second, and when present third, outlets comprises a concentration of CHO cells at the centre of the curvilinear microchannel, by width, that is greater than the concentration of CHO cells adjacent to the radial inner and radial outer wall of the curvilinear microchannel.

6. The method of any one of the preceding claims, wherein the suspension/fluid flowing through the first, second, and when present the third outlet, each have a different CHO cell density to that of the suspension/fluid supplied to the microfluidic device.

7. The method of any one of the preceding claims, wherein the Reynolds number (Re) of the flow through the microfluidic device is from 20 to 330, optionally from 120 to 330, more optionally from 120 to 230.

8. The method of any one of the preceding claims, wherein the first set of particles have an average particle size larger than the second set of particles.

9. The method of Claim 1 or 8, wherein the first set of particles comprises CHO cell clusters, optionally wherein the CHO cell clusters have a diameter of greater than 30 pm.

10. The method of any one of Claims 1 , 8 or 9, wherein the second set of particles is substantially free of CHO cell clusters having a diameter of greater than 30 pm, optionally wherein less than 10 wt%, such as less than 5 wt%, less than 3 wt%, less than 2 wt%, less than 1 wt% or less than 0.1 wt% of the mass of CHO cells in the second set of particles is composed of CHO cell clusters having a diameter of greater than 30 pm.

11 . The method of any one of the preceding claims, wherein the height of the radial inner side of the curvilinear microchannel is greater than the height of the radial outer side of the curvilinear microchannel.

12. The method of Claim 11 , wherein the ratio of the height of the radial inner side of the curvilinear microchannel to the height of the radial outer side of the curvilinear microchannel is from about 1.1 :1 to about 2.9:1 , optionally from about 1.1 :1 to about 2.5:1 , such as about 1.2:1 to about 2.1 :1.

13. The method of any one of Claims 1 to 10, wherein the height of the radial outer side of the curvilinear microchannel is greater than the height of the radial inner side of the curvilinear microchannel.

14. The method of Claim 13, wherein the ratio of the height of the radial outer side of the curvilinear microchannel to the height of the radial inner side of the curvilinear microchannel is from about 1.1 :1 to about 1.7:1 , such as about 1 .3:1 to about 1.5:1 .

15. The method of any one of the preceding claims, wherein the ratio W:H is from about 2:1 to about 10:1 , such as about 3:1 to about 5:1 ; where H is the greater of the height of the radial inner side of the curvilinear microchannel and the height of the radial outer side of the curvilinear microchannel; and W is the width of the curvilinear microchannel

16. The method of Claim 15, wherein the ratio W:H is from about 4:1 to about 9:1 .

17. The method of any one of the preceding claims, wherein the width of the curvilinear microchannel is substantially uniform along its length.

18. The method of any one of the preceding claims, wherein:

19. The method of any one of the preceding claims, wherein the curvilinear microchannel has a cross-sectional area of from about 0.5 x 10⁶ to about 3 x 10⁶ pm², such as about 1.5 x 10⁶ to about 3 x 10⁶ pm².

20. The method of any one of the preceding claims, wherein the curvilinear microchannel is a spiral microchannel, optionally wherein the spiral microchannel comprises at least 2 loops, such as at least 4 loops, for example at least 7 loops.

21. The method of any one of the preceding claims, wherein the curvilinear microchannel has a radius of curvature of from about 2.5 mm to about 25 mm.

22. The method of any one of the preceding claims, wherein the top and/or bottom side of the curvilinear microchannel has a linear cross section.

23. The method of any one of the preceding claims, wherein the flow rate of the suspension/fluid through the curvilinear microchannel is from about 1 to about 30 mL/min, such as about 5 to about 15 mL/min, for example about 10 mL/min.

24. The method of any one of the preceding claims, wherein the flow rate of the suspension/fluid through the curvilinear microchannel is from about 0.5 to about 10 mL/min, optionally from about 3 to about 10 mL/min, more optionally from about 4 to about 8 mL/min.

25. The method of any one of the preceding claims, wherein the suspension/fluid is supplied to the inlet of the microfluidic device by a pump configured to pump the suspension/fluid through the microfluidic device.

26. The method of any one of the preceding claims, which is performed in a bioreactor, where the suspension/fluid comprising single CHO cells and CHO clusters is taken from the bioreactor, optionally wherein the output from the first outlet is removed from the bioreactor and/or the output from the second outlet is re-supplied into the bioreactor.

27. The method of any one of the preceding claims, wherein the method comprises supplying the suspension/fluid to two or more microfluidic devices arranged in series, where an outlet of a first microfluidic device is connected to an inlet of a second microfluidic device.

28. The method of Claim 24, wherein the height of the radial inner side of the curvilinear microchannel of the first microfluidic device is greater than the height of the radial outer side of the curvilinear microchannel of the first microfluidic device, and the height of the radial inner side of the curvilinear microchannel of the second microfluidic device is lower than the height of the radial outer side of the curvilinear microchannel of the second microfluidic device.

29. The method of Claim 24 or 25, wherein the flow rate through the first microfluidic device is greater than the flow rate through the second microfluidic device, optionally wherein the flow rate through the first and second microfluidic devices are both as defined in Claim 23 or 24.

30. The method of any one of the preceding claims, wherein: