SYSTEM FOR MEMBRANE CHROMATOGRAPHY

Abstract

A chromatography system is provided. The chromatography system is configured to process a feed fluid containing a plurality of components, wherein at least one component of the plurality of components of the feed fluid is a target component. The chromatography system comprises: a flow path comprising a plurality of fluid control components configured to control a fluid flow; a stationary phase, wherein the stationary phase is at least one membrane adsorber connected to the flow path and the stationary phase is configured to isolate the target component. The flow path is configured such that harvesting of the target component is optimized.

Claims

1. A chromatography system configured to process a feed fluid containing a plurality of components, wherein at least one component of the plurality of components of the feed fluid is a target component and wherein the chromatography system comprises: a flow path comprising a plurality of fluid control components configured to control a fluid flow; and a stationary phase, wherein the stationary phase is at least one membrane adsorber connected to the flow path and the stationary phase is configured to isolate the target component; wherein the flow path is configured such that harvesting of the target component is optimized.

2. The chromatography system of claim 1, wherein the plurality of fluid control components comprises: a first outlet valve connected to the at least one membrane adsorber and configured to be connected to a target component collection vessel; and a second outlet valve connected to the at least one membrane adsorber and configured to be connected to a waste collection vessel; wherein the first outlet valve and the second outlet valve have a switching time of less than about 3 seconds, preferably less than about 1 second, most preferably equal to about 0.5 seconds.

3. The chromatography system of claim 1, wherein the plurality of fluid control components comprises: a first outlet valve connected to the at least one membrane adsorber and configured to be connected to a target component collection vessel; a second outlet valve connected to the at least one membrane adsorber and configured to be connected to a waste collection vessel; and a check valve positioned after the second outlet valve; wherein the second outlet valve has a switching time of about 3 seconds or greater than 3 seconds.

4. The chromatography system of claim 1, wherein the plurality of fluid control components further comprises: a first inlet valve configured to be connected to a feed fluid supply; and a second inlet valve configured to be connected to a buffer supply; wherein the first inlet valve and the second inlet valve have a switching time of less than about 3 seconds, preferably less than about 1 second, most preferably equal to about 0.5 seconds.

5. The chromatography system of claim 1, wherein the plurality of fluid control components further comprises: a first inlet valve configured to be connected to a feed fluid supply; a second inlet valve configured to be connected to a buffer supply; and at least one inlet check valve positioned after the first inlet valve and the second inlet valve; and wherein the first inlet valve and the second inlet valve have a switching time of about 3 seconds or greater than 3 seconds.

6. The chromatography system of claim 1, wherein the plurality of fluid control components further comprises a filter configured to filter only the feed fluid.

7. The chromatography system of claim 6, wherein the at least one membrane adsorber has a first pore diameter and the filter has a second pore diameter, the second pore diameter being less than the first pore diameter.

8. The chromatography system of claim 1, wherein the plurality of fluid control components further comprises an absorption detector positioned after the at least one membrane adsorber, and wherein a sampling rate of the absorption detector is less than about 0.7 s, preferably less than about 0.5 s, more preferably less than or equal to about 0.3 s.

9. The chromatography system of claim 1, wherein the plurality of fluid control components further comprises an absorption detector positioned before the at least one membrane adsorber.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0101] Details of exemplary embodiments are set forth below with reference to the exemplary drawings. Other features will be apparent from the description, the drawings, and from the claims. It should be understood, however, that even though embodiments are separately described, single features of different embodiments may be combined to further embodiments.

[0102] FIG. 1 shows a conceptual representation of an exemplary chromatography system.

[0103] FIG. 2 shows a conceptual representation of an inlet line in a flow path.

[0104] FIG. 3 shows plots of the maximum volume of the membrane adsorber for which there is no loss of product as a function of some features of the flow path.

[0105] FIG. 4 shows an optimisation routine for determining the optimal switching time.

[0106] FIG. 5 shows a plot of a UV detection signal vs volume for different sampling rates of a UV filter in the chromatography system.

[0107] FIG. 6 shows an enlarged plot of a UV detection signal vs volume for different sampling rates of a UV filter in the chromatography system.

[0108] FIG. 7 shows a plot of a UV detection signal vs time for different configurations of a filter in the chromatography system.

[0109] FIG. 8 shows a plot of a conductivity signal vs volume for different configurations of a filter in the chromatography system.

[0110] FIG. 9 shows a plot of normalized area vs normalized volume for different configurations of a filter in the chromatography system.

[0111] FIG. 10 shows a plot of a UV detection signal and a conductivity signal vs time for different configurations of a filter in the chromatography system.

[0112] FIG. 11 shows a schematic representation of an exemplary chromatography system.

[0113] FIG. 12 shows another schematic representation of an exemplary chromatography system.

DETAILED DESCRIPTION

[0114] In the following, a detailed description of examples will be given with reference to the drawings. It should be understood that various modifications to the examples may be made. Unless explicitly indicated otherwise, elements of one example may be combined and used in other examples to form new examples.

[0115] FIG. 1 shows a conceptual representation of an exemplary chromatography system 100. The chromatography system 100 comprises a flow path 110 and at least one membrane adsorber 200 as stationary phase. The flow path 110 comprises conduits such as pipes and/or tubing in which fluids can flow and it comprises a plurality of fluid control components configured to regulate the flow of the fluids. The chromatography system 100 comprises a control system (not shown) configured to manage at least part of the fluid control components. The chromatography system 100 may in particular be suitable for membrane chromatography in bind and elute mode.

[0116] The dashed elements in FIGS. 1 and 2 are optional. The flow path 110 may comprise one or more inlet lines at the beginning, wherein, in case of a plurality of inlet lines, the inlet lines conjoin into a main line. An exemplary inlet line, as shown in more detail in FIG. 2, comprises at least one inlet valve 120 configured to lead an input fluid, such as the feed fluid, a buffer or a washing fluid, into the flow path. In case the flow path 110 comprises only one inlet line, the inlet line comprises at least two inlet valves 120, one dedicated to the feed fluid and the other for other input fluids. Each inlet valve 120 is configured to be connected to at least one input fluid supply.

[0117] The inlet line further comprises a pump 125 and, between the inlet valve(s) 120 and the pump 125, an air sensor 130 configured to detect air in the inlet tubing may be installed. After the pump 125, the inlet line may comprise a check valve 140, e.g. if the switching time of the inlet valve(s) 120 is about 3 seconds or greater than about 3 seconds. After the check valve 140, other sensors 130 may be installed, such as a pressure sensor and a flowmeter, for monitoring the inlet line.

[0118] If the inlet line is a line dedicated for the feed fluid, the feed inlet line may comprise, after the sensors 130 following the pump 125, a filter 150 configured to filter the feed fluid in order to e.g. eliminate some particles. In particular, the filter may be directly inserted in the feed inlet inline, i.e. without valves. Alternatively, the filter 150 may be positioned in the main line. An analysis of the effects of the filter position is provided below with reference to FIGS. 7 to 10.

[0119] If the flow path comprises a plurality of inlet lines, each inlet line may comprise at the end, i.e. before joining the main line with the other inlet lines, a mix valve 145. The provision of mix valves separates the different fluids from one another and prevents back-mixing, as also discussed with reference to FIG. 10 below.

[0120] Going back to FIG. 1, the flow path 110 comprises a main line, after the inlet line(s), to which the membrane adsorber(s) 200 are connected. In one example, a single membrane adsorber 200 or a stack of membrane adsorbers 200 on top of each other may be connected to the flow path 110. In another example, two membrane adsorbers 200 or two stacks of membrane adsorbers 200 may be connected in parallel to the flow path 110. The use of two membrane adsorbers 200 or two stacks in parallel may allow a combination of capture and flow-through modes or an augmentation of the membrane volume and, thus, of the binding capacity. The term membrane volume refers to the volume of a single membrane adsorber 200 or a stack of membrane adsorbers 200 considering the porosity. For example, a membrane volume of 1 L may be the result of 200 mL of membrane layers and 800 mL of porosity.

[0121] The membrane adsorber(s) 200 may be connected to the flow path 110 by means of (membrane) valves. The main line may branch into a route on which there is no membrane adsorber 200 and one or two routes to which membrane adsorber(s) 200 can be connected.

[0122] The main line of the flow path 110 may comprise a filter 150 upstream of the membrane adsorber(s) 200, if the filter 150 is not positioned in the feed inlet line. The filter 150 may be connected to the flow path by means of (filter) valves. Thus, the main line may have two alternative branches, one with the filter 150 and one without the filter 150.

[0123] The main line of the flow path 110 may comprise one or more sensors 160 before the membrane adsorber(s) 200 (and after the filter 150, if present). In particular, a UV sensor may be positioned before the membrane adsorber(s) 200 to provide a monitoring function and to enable an adaptive control of the system. Other sensors 160 may include a pressure sensor, a conductivity sensor and a pH sensor.

[0124] The main line of the flow path 110 comprises one or more sensors 170 after the membrane adsorber(s) 200. In particular, at least a UV sensor 170 is positioned between the membrane adsorber(s) 200 and the outlet valves, wherein the UV sensor is configured to detect whether the fluid coming from the membrane adsorber(s) 200 contains the target component and should, thus, be directed towards a product collection vessel or towards other outlets, e.g. waste. Other sensors 170 may include a pressure sensor, a conductivity sensor and a pH sensor.

[0125] A signal generated by the UV sensor 170 is used by the control system to control the outlet valves positioned at the end of the flow path after the UV sensor 170. A sampling rate of the UV sensor 170 may be less than about 0.7 s, preferably less than about 0.5 s, more preferably less than or equal to about 0.3 s. A discussion of the sampling rate is given with reference to FIGS. 5 and 6 below.

[0126] The flow path 110 comprises at least two outlet vales 180 and 185 and optionally additional outlet vales, wherein each outlet valve is configured to be connected to a collection vessel. Outlet valve 180 may be connected to a target component collection vessel (and, thus, be denoted as target component outlet valve) while outlet valve 185 may be connected to a waste collection vessel (and, thus, be denoted as waste outlet valve).

[0127] The flow path 110 may comprise a check valve 195 after waste outlet valve 185, if the switching time of the waste outlet valve 185 is about 3 seconds or greater than about 3 seconds. In some examples, the flow path 110 may comprise a check valve after each outlet valve, respectively, if the switching time of the corresponding outlet valve is about 3 seconds or greater than about 3 seconds.

[0128] For the design of the chromatography system the portion of the flow path between the (post-membrane) UV sensor 170 and the output valves is of particular relevance. The maximum membrane volume V.sub.MA that can be operated without loss is coupled with the volumetric flow rate and, for fast non-kinetically limited stationary phases, the chromatography system may be designed accordingly. This relationship can be derived from the following equations (1) to (4):

[00001]$\begin{array}{cc}\stackrel{.}{V}=u.Math.A& \left(1\right)\\ A=\frac{}{4}.Math.d^2& \left(2\right)\\ V=A.Math.L& \left(3\right)\\ u=\frac{L}{t_{\mathrm{tot}}}& \left(4\right)\end{array}$

[0129] where {dot over (V)} is the maximal volumetric flow rate with no loss, d is the diameter and L is the length of the pipe between the UV sensor 170 and the output valves, and t.sub.tot is the total signal transmission time. The total signal transmission time consists of all time delays in signal transmission from the moment the UV sensor detects the passage of the target component until the signal is executed, i.e. until the target component outlet valve 180 is opened. Equation (5) shows an exemplary decomposition of t.sub.tot in time needed by the control system for the signal transmission, sensor sampling rate and valve switching time:

t.sub.tot=t.sub.cont+t.sub.samp.rate+t.sub.switch(5)

[0130] The maximal volumetric flow rate {dot over (V)} can be expressed as number of membrane volumes per unit of time {dot over (V)}=MV.Math.V.sub.MA.

[0131] FIG. 3 shows the maximum membrane volume for which there is no loss of product as a function of the pipe diameter d, the pipe length L and the target component outlet valve switching time t.sub.switch, for a volumetric flow rate of 5 membrane volumes per minute (MV=5/min).

[0132] With increasing valve switching time and decreasing pipe length, the maximum possible volume V.sub.MA is reduced. With increasing pipe diameter, higher maximum volumes of the stationary phase can be achieved with shorter tube lengths due to the reduced flow velocity.

[0133] Assuming 0.25 s for t.sub.cont at MV=5/min and a pipe length of 0.25 m, the following table shows the maximum stationary phase volume V.sub.MA before product loss occurs as a function of t.sub.tot. As the total signal transmission time increases, the usable stationary phase volume and, thus, the operating range of the system without product loss are reduced.

TABLE-US-00002 t.sub.tot [s] t.sub.switch [s] t.sub.samp.Math.rate [s] V.sub.MA [mL] 0.25 0 0 855 0.55 0 0.3 388 1.25 0 1 171 1.05 0.5 0.3 204 3.55 3 0.3 60 1.75 0.5 1 122 4.25 3 1 50

[0134] From all the considerations above it can be seen that there is an interplay between the variables t.sub.tot, L, d, MV when it comes to minimizing or eliminating product loss for a given membrane volume. Accordingly, an optimisation routine may be applied, such as

[00002]$\begin{array}{cc}\mathrm{Max}\left(V_{\mathrm{rel}.Math.\mathrm{Ma}}.Math.\frac{1}{V_{\mathrm{rel}}}\right)=f\left(t_{\mathrm{tot}},L,d,\mathrm{MV}\right)& \left(6\right)\end{array}$

which can then be used to solve for the optimum, for example by multiple linear regression or other systems of equations. V.sub.rel,MA is the membrane volume and V.sub.rel is the dead volume of the flow path or of the whole system (e.g. including the empty volume of a capsule housing the membrane adsorber). When their ratio is maximised, the result of the chromatographic process is mostly influenced by the characteristics of the chromatographic stationary phase, e.g. membrane adsorber. The smaller their ratio is, the stronger is the influence of the flow path.

[0135] FIG. 4 shows an optimisation routine for determining the optimal switching time with respect to the pipe length/diameter and t.sub.tot and thus the applicable stationary phase volume.

[0136] As discussed above, the sampling rate plays a role, in combination with other parameters, in optimizing the system to avoid product loss. FIGS. 5 and 6 illustrate the effect of the sampling rate alone on the performance of the chromatography system.

[0137] FIG. 5 shows a plot of a UV detection signal vs volume for different sampling rates of a UV filter in the chromatography system.

[0138] As mentioned before, the presence of the product/target molecule downstream the membrane adsorber is usually detected by checking whether a given condition (valve switching condition) is satisfied, i.e. that the absorption at a defined wavelength (e.g. 280 nm) is above a given threshold, e.g. 0.05 AU. As long as the absorption detected e.g. by a UV sensor is above the threshold, the product is collected by maintaining the target component outlet valve 180 open.

[0139] FIG. 5 shows an elution peak for BSA obtained with Sartobind Q and shows the cut points on the elution curve for two different scanning rate values, namely 1 s and 0.3 s. The cut points are the points at which the UV sensor realizes that the absorption has crossed the threshold.

[0140] For different scanning rate values the cut points occur at different times/volumes and the difference is clearly shown in the enlarged FIG. 6. Although the condition is met already at about 2525 mL, the UV sensor detects it with a delay in both cases. However, for a flow rate of 45 L/h, with the faster sampling rate the detection occurs after less than 4 mL, while with the sampling rate of 1.0 s it takes place after more than 12 mL. Higher flow rates would lead to higher volume distances between the cut points.

[0141] The hatched regions indicate the volume interval during which the product is collected. Considering the difference between the integral of the whole elution peak and of each hatched fraction thereof, respectively, it is possible to calculate the product loss.

TABLE-US-00003 Sampling Rate [s] Loss [%] 0.3 0.1 1.0 1.1

[0142] Accordingly, the reduction of the scanning rate results in the reduction of product loss.

[0143] Besides the sampling rate, another parameter that has been considered in combination with others with reference to FIGS. 3 and 4 is the switching time of the target component outlet valve 180. The effect of the valve switching time, taken alone, on the performance of the chromatography system is illustrated in the following with reference to five tests.

[0144] A product elution is simulated in tests 1, 2 and 3 with water and water/acetone (2-5% v/v), since acetone, like proteins, absorbs light at a wavelength of 280 nm and is therefore suitable as a model. In test 4 and 5, the analysis is performed using a Sartobind Q loaded with 1 L bovine serum albumin (BSA) (c=3 g/L) eluted with 0.5 M NaCl. The valve switching condition is 0.1 AU for all experiments. Each test was performed at least 3 times and appropriate fractions were drawn and analysed, the results are shown in the following table.

TABLE-US-00004 Test Mean Standard deviation Relative deviation number [AU] [g/L] [AU] [g/L] [%] 1 0.161 AU 0.054 33.4 2 0.890 AU 0.028 3.1 3 0.708 AU 0.021 3.0 4 9.7 g/L 0.400 4.0 5 9.7 g/L 0.300 3.3

[0145] In tests 1 and 2 the switching time of outlet valves 180 and 185 was set to about 3 s. In test 2, a check valve 195 was positioned after the waste outlet valve 185. In tests 3, 4 and the switching time of outlet valves 180 and 185 was set to about 0.5 s.

[0146] In comparison with tests 2 and 3, test 1 shows a significantly lower mean signal strength of the collected fractions. Furthermore, the deviation among the extracted fractions is with 33.4% the highest of all tests. The implementation of the check valve 185 leads to a significantly higher concentration of 0.890 AU and a relative deviation of 3.1%, which is significantly lower with respect to test 1. A comparable performance with test 2 is seen for tests 3, 4 and 5.

[0147] Therefore, the valve switching times have a considerable influence on the reproducibility and product concentration, wherein lower switching times are better. However, the valve switching time should not be too low, i.e. too close to 0 s, due to safety aspects such as pressure development in the system when pumping liquids at high volumetric flows. Hence, a valve switching time of less than about 3 seconds, preferably less than about 1 second, most preferably equal to about 0.5 seconds may be selected in order to reduce back-mixing and product loss.

[0148] Accordingly, the outlet valves 180 and 185 as well as optional additional outlet valves of the system 100 are set to have a switching time of less than about 3 seconds, preferably less than about 1 second, most preferably equal to about 0.5 seconds. Alternatively, the outlet valves 180 and 185 may have a switching time of about 3 seconds or more and at least the waste outlet valve 185 may have a check valve 195 positioned after it. If additional outlet valves are present, they may also have a check valve positioned thereafter. Optionally, also the target component outlet valve 180 may have a corresponding check valve 190.

[0149] Further, the input valves 120 may be controlled to have a switching time of less than about 3 seconds, preferably less than about 1 second, most preferably equal to about 0.5 seconds. Alternatively, the input valves 120 may have a switching time of about 3 seconds or more and each inlet line may comprise a check valve 140. The same concept applies to all valves in the flow path 110 of system 100.

[0150] Another aspect that may be considered in the design of the system is the position of the filter 150, if present. Depending on the position of the pre-filter, there are differences in the back-mixing, which are discussed below.

[0151] FIG. 7 shows a plot of a UV detection signal vs time for different configurations of a filter in the chromatography system. The three configurations are as follows: [0152] A) Filter connected to the main line by means of filter valves with a switching time of about 3 s, flow-through of feed fluid only (dotted line) [0153] B) Filter directly inserted in the main line, flow-through of all fluids (solid line) [0154] C) Filter directly inserted in the feed fluid inlet line (dashed line)

[0155] The same test procedure is performed with each configuration, wherein the test procedure is structured as follows: equilibration with water, loading with a 2-5% (v/v) water/acetone mixture, washing with water, a factitious elution and regeneration with water. The elution is carried out by gradually increasing the water/acetone mixture until a signal of 0.2 AU is reached, then regeneration is initiated.

[0156] Configuration A shows a dip in the signal at about 0.8 minutes. In the further course of the test the elution peak is identified at minute 3 and at minute 3.7 a further peak is identified. The drop in concentration during the loading process can be explained by back-mixing at the filter between feed solution and water, due to the volume of piping between the pumps and the filter as well as the switching time of the filter valves. This mixing leads to unwanted dynamic concentration profiles in the loading step, which negatively influence the binding characteristics of the stationary phase.

[0157] In the case of configuration B, the signal does not exhibit a reduction during loading but it does not reach a constant value. The elution peak is clearly wider than the one of configuration A and a second peak is visible also here. Since all fluids go through the filter, each fluid mixes with the residuals of the previously filtered fluid(s). This intermixture leads to a change in concentration that negatively affects the performance of the system.

[0158] For configuration C, the signal shows a steep rise in the loading phase and a sharp peak in the elution phase. A second peak is identified also here. All considered, this configuration shows the best results for fluid dynamics/back-mixing: the signal is stable at all times and shows a narrow elution peak.

[0159] While configuration C has the lowest back-mixing volume, configuration A has the option to have inline dilution in the system, while also having a narrow peak. The performance of configuration A can be improved by reducing the switching time of the filter valves, as shown in FIG. 8.

[0160] FIG. 8 shows a plot of a conductivity signal vs volume for different configurations of a filter in the chromatography system. In particular, the solid line represents configuration C, in which the filter is installed directly in the feed inlet line (also referred to as inline filter). The dashed line represents configuration A, in which the filter is connected to the main line by two filter valves (also referred to as online filter), with the modification that the filter valves have a switching time of about 0.5 s. Finally the dotted line also represents a configuration with an online filter and switching time of about 0.5 s.

[0161] All the curves have been obtained by sending different water and water/acetone mixtures through the respective filter positions at 45 L/h. It can be seen that the performance of the online filter with a valve switching time of 0.5 s is comparable to that of the inline filter. This is also visible in FIG. 9, which shows a plot of conductivity normalized area vs signal area centroid normalized volume for the inline filter and the online filter.

[0162] The less satisfactory performance of configuration B with respect to configuration C can be explained theoretically by using the equilibrium dispersive model expressed by equation 7 below, wherein c.sub.i is the concentration of a component in the feed fluid, u.sub.int is the linear velocity of the feed fluid and D.sub.ax is the coefficient of axial dispersion, which is the sum of the axial molecular diffusion and the eddy diffusion contribution.

[0163] In the general rate or equilibrium dispersive model the concentration changes with the time and here is calculated by the concentration change over the length (i.e. the dimension in the flow direction). Furthermore, the concentration change with the time is divided into convective and diffusive/dispersive mass transfer. The convective term describes the concentration flow to the next length section by the linear velocity and the length change. The diffusive/dispersive mass transfer is described by the axial dispersion coefficient. The axial dispersion dimension/effect is described by the value of the axial dispersion coefficient and the change of concentration through the cross sectional area represented by the second derivative of the concentration with respect to the length, comparable to the second Fick's law. The second derivative of the concentration with respect to the length represents the back-mixing.

[0164] In other words, the change in concentration over time results from the convective transport via the linear velocity as well as the back-mixing via the axial dispersion coefficient. More precisely, the longer the volume is flowed by a fluid, the higher is the influence of the back-mixing. The locally-considered concentration change .sup.2c.sub.i increases with the increase of the sum of the temporal concentration change over the quotient of the length and axial dispersion coefficient, as shown by equation 8.

[00003]$\begin{array}{cc}\frac{c_i}{t}=-u_{\mathrm{int}}.Math.\frac{c_i}{x}+\underset{\underset{1}{}}{D_{\mathrm{ax}}.Math.\frac{{}^2{c}_i}{x^2}}& \left(7\right)\\ \frac{c_i}{t}.Math.\frac{x^2}{D_{\mathrm{ax}}}={}^2{c}_i& \left(8\right)\end{array}$ [0165] c=concentration of i.sup.th component [g/L/M] [0166] t=time [s] [0167] u=linear velocity [mm/s] [0168] x=longitudinal coordinate [mm] [0169] D.sub.ax=axial dispersion coefficient [mm.sup.2/s]

[0170] If the filter is permanently exclusively flowed through by the feed, as in configuration C, the back-mixing in the system is significantly reduced, as the concentration change c.sub.i/t is 0 within the filter.

[0171] If the filter is not installed directly in the inlet line used exclusively for the feed, it should be only flowed through by the feed. Further, it should be permanently, i.e. during the whole cycle, filled with feed. Therefore, the switching time of the filter valves would have to be adjusted by the volume that the feed travels back to the filter valve, in order to avoid concentration gradients.

[0172] In all configurations A, B and C a second elution peak is detected, as shown in FIG. 7. FIG. 10 shows the conductivity signal (dashed line) in addition to the UV signal (solid line) for configuration A. The conductivity signal features a small peak at the beginning of the loading, a salt signal, which reduces the binding capacity of the stationary phase. This undesirable behaviour occurs due to the merging of a plurality of inlet lines without efficient prevention of back mixing.

[0173] The provision of a mix valve 145 on each inlet line creates a dedicated mixing point and separates the different media from each other, so that the concentration gradient is 0 up to the mixing point, which means that back-mixing cannot take place (see equation 7). In the case of an online filter, the presence of mix valves also allows a reduction of the distance between the point at which the inlet lines meet and the position of the filter.

[0174] A further measure to reduce back-mixing concerns the general structure of the flow path from inlet to outlet. A flow path in which bends and turns are minimized, or, in other words, a flow path that is as straight as possible reduces the dead volume of the flow path and, thus, back-mixing.

[0175] FIGS. 11 and 12 show two examples of a chromatography system in which one or more of the measures for optimizing the chromatographic process as illustrated heretofore are implemented.

[0176] FIG. 11 shows an exemplary implementation of a chromatography system 900. The system 900 has three inlet lines, each one comprising a plurality of inlet valves 901/903/905, an air sensor 907/909/911, a pump 913/915/917, a check valve 919/921/923, a pressure sensor 925/927/929 and a flowmeter 931/933/935. One inlet line is for the feed, another one is for water to perform inline dilution, and the last one is for the buffers.

[0177] After the three inlet lines are merged, a filter 940 is provided along the flow path. In particular, the flow path is provided with two filter valves 942, 944 for connecting the filter 940. The flow path further comprises a filter bypass valve 946 and a discharge valve 948, which provide alternative routes to an incoming fluid.

[0178] Subsequently two membrane adsorbers (or two stacks) 960, 961 are connected to the flow path, each by means of two membrane valves 962, 963 and 964, 965, respectively. The volume of the each membrane adsorber/each stack is 150 mL. Sensor sets 950 and 970 for measuring pressure, conductivity, pH and absorption are implemented before and after the membrane adsorbers 960, 961. The flow path further comprises membrane bypass valve 966.

[0179] Finally, four outlet valves 980, 985, 990 and 995 are installed for the discharge of waste, the final product and various process intermediates.

[0180] All the valves in system 900 are controlled to have a switching time of less than about 3 seconds, excluding the plurality of inlet valves 901, 903, 905, after which, however, check valves 919, 921, 923 are implemented, respectively.

[0181] As mentioned, due to the smaller volume of the stationary phase compared to a traditional column chromatography and the associated lower binding capacity per cycle, a higher number of membrane chromatography cycles is required to process a given volume of feed. However, each cycle is shorter, as can be seen from the following table for a protein A chromatography performed with system 900.

TABLE-US-00005 Volume Flow rate Residence Duration Step [MV] [MV/min] time [min] [min] (Re)Equilibration 12 5 0.2 2.4 Load 6.7 5 0.2 1.3 Wash 5 5 0.2 1 Elution ~2 5 0.2 0.4 Regeneration 10 5 0.2 2 CIP 1 5 0.2 0.2 Total 36.7 N/A N/A 7

[0182] The above values refer to a cycle of membrane chromatography in bind and elute mode with fixed transition from one step to the following. The volume for each step is expressed in units of the membrane volume, which is 150 mL. The residence time of the protein-containing solution on the membrane is 20 seconds, because convection is primarily responsible for mass transport, leading to a more efficient adsorption with respect to the diffusion mechanism that is predominant in resins. The dynamic binding time is about 20 g/L.

[0183] The cycle of membrane chromatography executed with system 900 of FIG. 9 lasts about 7 minutes, while the corresponding cycle with column chromatography takes more than 4 hours, as illustrated previously.

[0184] FIG. 12 shows another exemplary implementation of a chromatography system 1000. The chromatography system 1000 comprises two inlet lines, each one comprising a plurality of inlet valves 1001/1003, an air sensor 1005/1007, a pump 1009/1011, a pressure sensor 1013/1015, a flowmeter 1017/1019 and a mix valve 1021/1023. One inlet line is for the feed and the other one is for the buffers.

[0185] After the two inlet lines are merged, a filter 1040 is provided along the flow path. In particular, the flow path is provided with two filter valves 1042, 1044 for connecting the filter 1040. The flow path further comprises a filter bypass valve 1046 and a discharge valve 1048, which provide alternative routes to an incoming fluid.

[0186] Subsequently two membrane adsorbers (or two stacks) 1060, 1061 are connected to the flow path, each by means of two membrane valves 1062, 1063 and 1064, 1065, respectively. The volume of the each membrane adsorber/each stack is 150 mL. Sensor sets 1050 and 1070 for measuring pressure, conductivity, pH and absorption are implemented before and after the membrane adsorbers 1060, 1061. The flow path further comprises membrane bypass valve 1066.

[0187] Finally, four outlet valves 1080, 1085, 1090 and 1095 are installed for the discharge of waste, the final product and various process intermediates.

[0188] All the valves in system 1000 are controlled to have a switching time of less than about 3 seconds.

[0189] The duration of each phase of a cycle for a protein A chromatography performed with system 1000 is reported in the following table:

TABLE-US-00006 Volume Flow rate Residence Duration Step [MV] [MV/min] time [min] [min] (Re)Equilibration 6.5 5 0.2 1.3 Load 6.7 5 0.2 1.3 Wash 1.8 5 0.2 0.4 Elution ~2 5 0.2 0.4 Regeneration 2 5 0.2 0.4 CIP 1 5 0.2 0.2 Sum 20 N/A N/A 4

[0190] The above values refer to a cycle of membrane chromatography in bind and elute mode with conditional transition from one step to the following, e.g. moving to the following step when a given UV absorption value or a conductivity value has been reached.

[0191] The cycle of the membrane chromatography executed with system 1000 of FIG. 10 lasts about 4 minutes, hence system 1000 is faster than system 900 shown in FIG. 9. Part of the reason is that, as explained before, the presence of the mix valves allows to reduce a length of the flow path between the inlet lines and the filter. Further, the conditional transition makes the cycle shorter.

[0192] The reduction in back-mixing achieved by the structural design and/or flow control measures discussed heretofore results in a low peak broadening of the elution fraction and thus a high product concentration in the eluate. Therefore, there is no accumulation of dilutions over many cycles and the chromatographic process is improved in terms of efficiency and quality.