CHROMATOGRAPHIC PURIFICATION METHOD

Abstract

A chromatographic purification method for the isolation of a desired product fraction from a mixture using 2 chromatographic columns, comprises, within one cycle to be carried out at least once, the following steps: a first batch step, wherein said columns are disconnected and a first column is loaded with feed and its outlet is directed to waste, and from a second column desired product is recovered and subsequently the second column is regenerated; a first interconnected step, wherein the outlet of the first column is connected to the inlet of the second column, the first column is loaded beyond its dynamic breakthrough capacity with feed, and the outlet of the second column is directed to waste, a second batch step analogous to the first batch step but with exchanged columns; and a second interconnected step, analogous to the first interconnected step but with exchanged columns.

Claims

1.-15. (canceled)

16. A chromatographic purification method for an isolation of target molecules from a feed consisting of the target molecules and impurities to be separated from the target molecules, the target molecules being adsorbed on an affinity chromatography material having immobilized ligands that bind specifically to the target molecules while letting the impurities pass by unaffected, the method using only two chromatographic columns consisting of a first column and a second column, the first column and the second column being loaded with the affinity chromatography material, the first column having a first column inlet and a first column outlet, the second column having a second column inlet and a second column outlet, the method comprising the following steps in order: a first batch step, performed during a batch timespan, wherein the two chromatographic columns are disconnected, the first column is loaded with the feed via the first column inlet at a first flow rate and the first column outlet is directed to a waste, and from the second column, the target molecules are recovered via the second column outlet based on the second column being subjected to a regeneration process; a first interconnected step, performed during an interconnected timespan, wherein the first column outlet is connected to the second column inlet, the first column thus being a first upstream column and the second column thus being a first downstream column, the first column has a dynamic breakthrough capacity of the target molecules beyond which the target molecules exit the first column via the first column outlet, the first column is loaded beyond the dynamic breakthrough capacity with the feed via the first column inlet, the second column outlet is directed to the waste, and wherein during a subsequent washing timespan, the first column outlet remains connected to the second column inlet, the first column and the second column are subjected to a subsequent washing process by washing with a subsequent-washing-solvent, which is the same as or different from the washing-solvent, and/or a subsequent-washing-buffer, which is the same as or different from the washing-buffer, and an eluate exiting the second column outlet is directed to the waste; a second batch step, wherein the first column performs all tasks of the second column in the first batch step, and the second column performs all tasks of the first column in the first batch step; and a second interconnected step, wherein the second column outlet is connected to the first column inlet, the first column performs all tasks of the second column in the first interconnected step and the second column performs all tasks of the first column in the first interconnected step, wherein at the first column outlet and the second column outlet, a concentration of the target molecules and/or a concentration of the impurities is detected using a detector.

17. The method according to claim 16, wherein the first column is loaded beyond the dynamic breakthrough capacity with the feed via the first column inlet at a second flow rate which is 1.5 to 4.0 times larger than the first flow rate.

18. The method according to claim 16, wherein said regeneration process comprises the following steps in order: (i) washing with a washing-solvent and/or a washing-buffer under a first condition that the target molecules are not released from the affinity chromatography material; (ii) eluting with an eluting-solvent and/or an eluting-buffer under a second condition that the target molecules are released from the affinity chromatography material; and at least one of the following steps (iii) and (iv): (iii) cleaning in place using a cleaning-solvent and/or a cleaning-buffer to release everything from the affinity chromatography material after the steps (i) and (ii); and (iv) equilibrating by using an equilibrating-solvent and/or an equilibrating-buffer.

19. The method according to claim 16, wherein said subsequent washing timespan, which is denoted as t.sub.wash,IC is determined according to the following equation:
t.sub.wash,IC=k.Math.V.sub.dead/Q.sub.wash,IC wherein V.sub.dead is a dead volume of a single column, Q.sub.wash,IC is a subsequent washing flow rate, and k is a rational number equal to or larger than 1.

20. The method according to claim 16, wherein said first batch step, first interconnected step, second batch step, and said second interconnected step form a cycle, and this cycle is carried out continuously and repeatedly until exchange of column material is necessary or until the feed flow is interrupted necessitating a stop or shutdown.

21. The method according to claim 16, wherein at least one of the second flow rate and the interconnected timespan is adapted such that at the end of the first and second interconnected steps the concentration of the target molecules at the first column outlet in the first interconnected step and the second column outlet in the second interconnected step, is in a range of 30-90% of the concentration of the target molecules at the first column inlet in the first interconnected step and the second column inlet in the second interconnected step, respectively, and wherein at least one of the second flow rate and the interconnected timespan is adapted such that at the end of the first and second interconnected steps the concentration of the target molecules at the second column outlet in the first interconnected step and the first column outlet in the second interconnected step, is below a breakthrough value of 5% of the concentration of the target molecules at the first column inlet in the first interconnected step and the second column inlet in the second interconnected step, respectively.

22. The method according to claim 16, wherein the batch timespan is an accumulated time required for the regeneration process in the first batch step or the second batch step.

23. The method according to claim 16, wherein before said first batch step a start-up step, is carried out, in which the second column outlet is connected to the first column inlet and the feed is loaded into the second column, and wherein in the start-up step, an amount of the feed that is loaded into the second column is larger than an amount of the feed that is loaded into the first column in the first interconnected step or the second column in the second interconnected step.

24. The method according to claim 16, wherein the target molecules are selected from the group consisting of chemical reaction products, chemical separation products, biochemical reaction products, biological products and a combination thereof.

25. A method for setting up a chromatographic process according to claim 16, wherein the batch timespan is an accumulated time required for the regeneration process in the first batch step or the second batch step, the first flow rate is set such that at the end of the first and second batch steps the target molecules are not eluted at the first column outlet and the second column outlet, respectively, and the second flow rate and the interconnected timespan are set such that at the end of the first and second interconnected steps the target molecules are not eluted at the second column outlet and the first column outlet, respectively.

26. The method according to claim 16, wherein the second flow rate and/or the interconnected timespan are adapted such that at the end of the first and second interconnected steps the concentration of the target molecules at the first column outlet in the first interconnected step and the second column outlet in the second interconnected step, is in a range of 30-90% of the concentration of the target molecules at the first column inlet in the first interconnected step and the second column inlet in the second interconnected step, respectively, and wherein the second flow rate and/or the interconnected timespan are adapted such that at the end of the first and second interconnected steps the concentration of the target molecules at the second column outlet in the first interconnected step and at the first column outlet in the second interconnected step, is below a breakthrough value of 2.5% of the concentration of the target molecules at the first column inlet in the first interconnected step and the second column inlet in the second interconnected step, respectively.

27. The method according to claim 16, wherein the affinity chromatography material of the two chromatographic columns is in a form of particles, beads, membranes or monoliths.

28. The method according to claim 16, wherein the detector is selected from the group consisting of an UV detector, a visible light detector, an IR detector, a fluorescence detector, a light scattering detector, a refractive index detector, a pH detector, a conductivity detector, an at-line HPLC detector, and a mass spectrometry detector.

29. The method according to claim 16 further comprising a shutdown step after the second interconnected step, wherein in the shutdown step, the two chromatographic columns are disconnected, and are subjected to the regeneration process.

30. The method according to claim 20, wherein said first batch step, first interconnected step, second batch step, and said second interconnected step form a cycle, and wherein the method further comprises a shutdown step after the last interconnected step of the last cycle, wherein in the shutdown step, the two chromatographic columns are disconnected, and are subjected to the regeneration process.

31. The method according to claim 20, wherein said first batch step, first interconnected step, second batch step, and said second interconnected step form a cycle, and wherein the method further comprises a shutdown step after the last batch step of the last cycle, wherein in the shutdown step, the chromatographic columns are disconnected, and the column which has not been regenerated in the last batch step of the last cycle is subjected to the regeneration process.

32. The method according to claim 16, wherein the target molecules are selected from the group consisting of chemical reaction products, chemical separation products, biochemical reaction products, biological products, and a combination thereof, and wherein the chemical reaction products and the biochemical reaction products are selected from the group consisting of natural products, metals, antibodies, antibody fragments, fusion proteins, recombinant glycoproteins, plasma proteins, a derivative thereof and a combination thereof.

33. The method according to claim 16, wherein at least one of the second flow rate and the interconnected timespan is adapted such that at the end of the first and the second interconnected steps the concentration of the target molecules at the first column outlet in the first interconnected step and the second column outlet in the second interconnected step is in a range of 30-90% of the concentration of the target molecules at the first column inlet in the first interconnected step and the second column inlet in the second interconnected step, respectively, and wherein at least one of the second flow rate and the interconnected timespan is adapted such that at the end of the first and second interconnected steps the concentration of the target molecules at second column outlet in the first interconnected step and the first column outlet in the second interconnected step, the concentration of the target molecules at the second column outlet in the first interconnected step and the first column outlet in the second interconnected step is below a breakthrough value of 0.25% of the concentration of the feed at the first column inlet in the first interconnected step and the second column inlet in the second interconnected step, respectively.

34. The method according to claim 16, wherein before said first batch step a start-up step is carried out, in which the second column outlet is connected to the first column inlet and the feed is loaded into the second column.

35. The method according to claim 16, wherein the first column is loaded beyond the dynamic breakthrough capacity with the feed via the first column inlet at a second flow rate which is at least 10% larger than the first flow rate in the batch step.

36. The method according to claim 16, wherein the first column is loaded beyond the dynamic breakthrough capacity with the feed via the first column inlet at a second flow rate which is at least 25% larger than the first flow rate in the batch step.

37. The method according to claim 16, wherein the first column is loaded beyond the dynamic breakthrough capacity with the feed via the first column inlet at a second flow rate which is 1.5-2.0 larger than the first flow rate in the batch step.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0154] Preferred embodiments of the invention are described in the following with reference to the drawings, which are for the purpose of illustrating the present preferred embodiments of the invention and not for the purpose of limiting the same. In the drawings,

[0155] FIG. 1 shows a schematic representation of a twin column countercurrent sequential loading process, where the process comprises an optional startup phase, a cyclic phase and a shutdown phase;

[0156] FIG. 2 shows an example of a UV signal recorded at the outlet of column 1 by detector UV1 (black solid line) for the steps in FIG. 1. The plateau value of the non-adsorbing impurities is indicated by horizontal dotted lines;

[0157] FIG. 3 shows breakthrough curves, wherein diamonds indicate the single column breakthrough curve at high flow rate, squares indicate the single column breakthrough curve at low flow rate, triangles indicate the breakthrough curve from two serially interconnected columns; the elution volumes EV.sub.1H,1, EV.sub.1L,1 EV.sub.1H,75, and EV.sub.2 are indicated by solid vertical lines; the feed concentration is indicated by a dashed horizontal line;

[0158] FIG. 4 shows the performance parameters yield, productivity, IgG concentration, and buffer consumption (average values) for the five cycles of the sequential countercurrent loading process of example 2;

[0159] FIG. 5 shows signals recorded during cycles 2, 3, 4 by the two UV detectors (UV1, UV2) positioned at the column outlets of a twin-column sequential loading chromatography process, wherein the interconnected phases are denoted with IC, the disconnected phases are denoted with B; the phase borders are represented by black dashed short vertical lines; the cycle borders are represented by bold dashed long vertical lines; the areas corresponding to the breakthrough from the upstream column are marked with A.sub.IC2U,n for UV2 (column 2 is upstream) and A.sub.IC1U,n for UV1 (column 1 is upstream), respectively, where n stands for the cycle number; for instance A.sub.IC1U,3 refers to the area corresponding to the breakthrough from column 1 in the 3rd cycle with column 1 being in the upstream position; the peak areas corresponding to the eluates from the columns that have been in the upstream position in the previous interconnected phases are marked with A.sub.peak2,n for UV2 (column 2 was upstream in the previous interconnected phase) and A.sub.peak1,n for UV1 (column 1 was upstream in the previous interconnected phase), respectively, where n stands for the cycle number; for instance A.sub.peak1,3 refers to the area corresponding to the product elution from column 1 in the 3rd cycle; the baselines used for the area determinations are indicated with thin dashed lines;

[0160] FIG. 6 shows the signals recorded during cycles 2 and 3 by the two UV detectors (UV1, UV2) positioned at the column outlets of a twin-column sequential loading chromatography process; the phase borders are represented by dashed short vertical lines; the cycle borders are represented by dashed long vertical lines; the areas corresponding to the breakthrough of the non-adsorbing impurities cannot be determined due to non-linearity of the detector signal; the peak areas corresponding to the product eluates from the columns that have been in the upstream position in the previous interconnected phases are marked with A.sub.peaki2,n for UV2 (column 2 was upstream in the previous interconnected phase) and A.sub.peak1,n for UV1 (column 1 was upstream in the previous interconnected phase), where n stands for the cycle number 2 or 3; for instance A.sub.peak2,3 refers to the area corresponding to the product elution from column 2 in the 3rd cycle; the baselines used for the area determinations are indicated with thin dashed lines;

[0161] FIG. 7 shows an overlay of signals from two runs of a twin-column sequential loading chromatography process recorded by the UV detector UV2 during one cycle (B1, IC1, B2, IC2); for the sake of simplicity the cycle index n is omitted; the signal of the first run (UV2-I, full line) corresponds to operating parameters with a load of 48 g/L, the signal of the second run (UV2-II, dotted line) corresponds to operating parameters with a load of 30 g/L (decreased feed flow rate in the disconnected state); the interconnected phases are denoted with IC, the disconnected phases are denoted with B; the phase borders are represented by black dashed vertical lines. The plateau value of the non-adsorbing impurities is indicated by a horizontal dashed-dotted line. The areas referred to in the example 8 description are indicated by arrows;

[0162] FIG. 8 shows the productivity (diamonds) and yield determined by the optimization method described in example 9 (full circles) and by offline HPLC analyses (empty circles) as a function of the load; and

[0163] FIG. 9 shows the peak area values A.sub.peak from UV1 (diamonds) and UV2 (squares) according to example 11 as a function of the load; the values corresponding to the three lowest loads (UV1) and the two lowest loads (UV2), respectively, were fitted using linear functions (equations displayed in figure); in addition, the yield determined by the optimization method described in example 11 (full circles) and by offline HPLC analyses (empty circles) is shown as a function of the load.

DESCRIPTION OF PREFERRED EMBODIMENTS

Example 1: Initial Determination of Operating Parameters for a Twin Column Countercurrent Sequential Loading Process

[0164] The initial operating parameters for a twin-column countercurrent sequential loading process for the purification of an IgG from clarified cell culture harvest using protein A affinity chromatography were determined based on the procedure outlined above. The breakthrough curves were recorded, fractionated and analyzed by offline Protein A analysis using a Poros A/20 column (Life technologies, USA) to determine the IgG concentrations. The concentration of IgG in the feed was 1.0 g/L. The columns were of 0.5 cm inner diameter and 5.0 cm length. A protocol for the elution and regeneration of a loaded column was developed including a wash step of 6 min, an elution step of 7 min, a cleaning step of 6 min, a first equilibration step of 3 min, and a second equilibration step of 3 min; all at a flow rate of 1 mL/min. Thus, the total duration of the elution and regeneration steps was t.sub.B=25 min.

[0165] A first breakthrough curve for a single column was recorded for a feed flow rate of 1 mL/min, a second breakthrough curve for a single column at a feed flow rate of 0.5 mL/min and a third breakthrough curve at a feed flow rate of 1 mL/min for two columns in series. With the load factor X=75%, and the safety factor Z=90%, the following elution volume values were obtained (see above descriptions and equations):

EV.sub.1H,1=19.9 mL, EV.sub.1L,1=23.4 mL, EV.sub.1H,75=48.4 mL and EV.sub.2=60.8 mL.

[0166] An overlay of the offline analysis results together with marks for the elution volume values is provided in FIG. 3. Since EV.sub.1H,75<EV.sub.2, EV.sub.Y was set equal to EV.sub.1H,75. The preload value PL was computed by integration approximation using a trapezoid rule to be PL=11.2 g/L and the target load value was determined to be TL=16.3 g/L. The feed flow rate in the disconnected state was determined to be 0.53 mL/min. Since this flow rate was larger than the flow rate used to determine the second breakthrough curve, the latter value was selected as feed flow rate, Q.sub.feed,B=0.50 mL/min. The duration of the interconnected phase of the sequential loading process was computed to be t.sub.IC=25 min and the duration of the startup phase was t.sub.startup=47.5 min. The feed flow rate in both cases was Q.sub.feed,startup=Q.sub.feed,IC=1 mL/min. The interconnected washing step duration was t.sub.wash,IC=1 min and the flow rate was Q.sub.wash,IC.

Example 2: Operation of a Twin Column Countercurrent Sequential Loading Process

[0167] A twin column countercurrent sequential loading process was used for the capture of an IgG monoclonal antibody from clarified cell culture harvest using a Protein A affinity stationary phase packed into two columns of 0.5 cm inner diameter and 5.0 cm length. The process was operated with the following parameters on Contichrom Lab-10 equipment from ChromaCon AG, Switzerland. The UV detection wavelength was 305 nm. The process was run using the operating parameters determined in example 1, which are summarized in table 1.

TABLE-US-00001 TABLE 1 Operating parameters for twin-column countercurrent sequential loading process of example 2 including the startup phase. Q.sub.feed indicates the feed flow rate, Q.sub.buffer indicates the flow rate of all other steps; t is the duration of the substeps of the disconnected phase B and the interconnected phase IC, respectively. The parameters for the final elution correspond to the B parameters except for the feed flow rate, which is zero during the final elution step. Q.sub.buffer Q.sub.feed Phase step [mL/ buffer [mL/ t [] [] min] [] min] [min] B wash 1.0 25 mM Phosphate, 0.5 6.0 pH 7.0 elute 1.0 25 mM Citrate, pH 3.0 0.5 6.0 clean 1.0 0.1M NaOH 0.5 7.0 equilibrate 1 1.0 25 mM Citrate, pH 3.0 0.5 3.0 equilibrate 2 1.0 25 mM Phosphate, 0.5 3.0 pH 7.0 IC feed 0 1.0 25.0 wash 1.0 25 mM Phosphate, 0.0 1.0 pH 7.0 startup feed 0 1.0 47.5

[0168] The operating parameters of the chromatographic process (such as feed flow rate and switch times) were not changed throughout the run. The feed concentration was artificially increased by approximately 0.01-0.02 g/L every cycle (see table 2) to simulate a subtle change in the upstream process.

[0169] After the initial startup phase the purification process was run continuously over five cycles before being eluted in the final elution phase. The effluents corresponding to the feed, the product fraction, the feed flow through fraction and the cleaning fraction were analyzed by offline protein A HPLC analysis in order to determine the IgG concentrations. From the volume of the fractions and the product concentrations, the product masses and performance parameters were computed.

[0170] The yield was calculated by comparing the product mass in the product fraction with the product mass in the feed fraction. The productivity was calculated by division of the product mass (obtained in one cycle) by the cycle duration and the total column volume (i.e. the volume of the two columns together). The buffer consumption was calculated by dividing the overall volume of buffers used within one cycle by the product mass in the product fraction of that cycle. The feed concentration and the performance parameters are summarized in table 2 and the evolution over time is shown in FIG. 4. The example shows that high yield and productivity values can be obtained using the described process despite subtle changes of the feed concentration.

TABLE-US-00002 TABLE 2 Feed concentration, IgG concentration, yield, productivity and buffer consumption of the sequential countercurrent loading process as average values for the five cycles of the process. feed pool buffer conc conc yield productivity consumption Cycle [g/L] [g/L] [%] [g/L/h] [L/g] 1 1.01 5.10 53.8 21.0 1.1 2 1.03 5.24 93.3 21.5 1.1 3 1.04 5.28 92.6 21.7 1.1 4 1.06 5.37 92.6 22.1 1.0 5 1.08 5.49 93.1 22.6 1.0

Example 3: Detector Comparison, Low Impurity Signals

[0171] A twin column countercurrent sequential loading process was operated with the following parameters on Contichrom Lab-10 equipment from ChromaCon AG, Switzerland. Two detector cells giving different signals for the same sample were used for demonstration. The operating parameters are summarized in Table 3.

TABLE-US-00003 TABLE 3 Operating parameters for twin-column countercurrent sequential loading process of example 3. Q.sub.feed indicates the feed flow rate, Q.sub.buffer indicates the flow rate of all other steps; t is the duration of the substeps of the disconnected phase B and the interconnected phase IC, respectively. The parameters for the final elution correspond to the B parameters except for the feed flow rate, which is zero during the final elution step. The UV detection wavelength was 300 nm. Q.sub.buffer Q.sub.feed Phase step [mL/ buffer [mL/ t [] [] min] [] min] [min] B wash 1.0 25 mM Phosphate, pH 7.0 0.2 6.0 elute 1.0 25 mM Citrate, pH 3.0 0.2 6.0 clean 1.0 0.1M NaOH 0.2 7.0 equilibrate 1.0 25 mM Citrate, pH 3.0 0.2 3.0 1 equilibrate 1.0 25 mM Phosphate, pH 7.0 0.2 3.0 2 IC feed 0 1.0 24.5 wash 1.0 25 mM Phosphate, pH 7.0 0.0 1.0 startup feed 0 1.0 45.0

[0172] The chromatograms from the UV detectors UV1 and UV2 located at the outlet of each column, recorded during cycles 2, 3 and 4 are shown in FIG. 5. The process was in cyclic steady state in cycles 3 and 4.

[0173] During the interconnected phases, the UV detectors record a signal that corresponds to the load flow-through of the upstream and the downstream column, respectively. The chromatograms of the process in cyclic steady state are explained referring to cycle 2 starting at 100 min.

[0174] In the following, the index n is used to describe the n-th cycle of the process. Thus, for instance, the area A.sub.peak2,3 corresponds to the elution peak area recorded for column 2 (by detector 2) in the 3.sup.rd cycle.

[0175] In the disconnected phase B the product was recovered from column 2. The disconnected phase typically includes at least one washing step prior to elution, and typically a strip and/or clean step followed by at least one re-equilibration step. The product peak is indicated by A.sub.peak2,2 (recorded by UV2). In parallel to the washing, elution, cleaning and re-equilibration of column 2, column 1 was loaded with feed at lower flow rate as the feed flow rate in the interconnected state. The signal UV1 corresponded to the non-adsorbing impurities that flowed through. The signal reached a constant plateau, indicating that no product was breaking through during the B phase.

[0176] In the subsequent interconnected phase IC, the columns were interconnected and column 1 was placed in the upstream position and continued to be loaded. The rising signal (UV1) recorded at the outlet of column 1 at the end of the interconnected phase indicated that product was breaking through (A.sub.IC1U,2). However, this product was not lost but re-adsorbed in the downstream column 2. While breakthrough was detected at column 1 (UV1), no breakthrough was detected at column 2 (UV2) indicating that the entire product was re-adsorbed. At the end of the interconnected phase B column 1 was washed with fresh buffer such that the product present in the liquid volume of the upstream column 1 was adsorbed in the downstream column 2.

[0177] In the following disconnected phase B the product was recovered from column 1 using the same protocol of washing, elution, strip and/or clean, re-equilibration steps that was used in the previous disconnected phase for column 2. The product peak is indicated by A.sub.peak1,2 (recorded by UV1). At the same time column 2 was loaded with feed at a lower flow rate as the feed flow rate in the interconnected state. The signal UV2 corresponded to the impurities that flowed through. Again, the signal reached a constant plateau, indicating that no product was breaking through.

[0178] In the subsequent interconnected phase IC, column 2 was continued to be loaded and the rising signal (UV2) at the end of the interconnected phase indicated that product was breaking through from column 2 (A.sub.IC2U,2). However, this product was not lost but re-adsorbed in the downstream column. While breakthrough was detected from column 2 (UV2), no breakthrough was detected at column 1 (UV1) indicating that the entire product was re-adsorbed. At the end of the interconnected phase B column 2 was washed with fresh buffer such that the product present in the liquid volume of column 2 was adsorbed in the downstream column 1.

[0179] After the interconnected phase had been completed, a new cycle was started.

[0180] In the process, before being washed and eluted in the disconnected phase, the respective columns are in the upstream positions in the interconnected phase and loaded such that product is breaking through into the downstream column where it is re-adsorbed (see example 2). The breakthrough in the interconnected phase is proportional to the areas designated with A.sub.IC2U,n (UV2, column 2) and A.sub.IC1U,n (for UV1, column 1) in FIG. 5, with n as cycle number.

[0181] In FIG. 5 furthermore the peak areas corresponding to the elution of the columns in the disconnected phase that have previously been in the upstream position in the interconnected phase are shown and designated A.sub.peak1,n (for UV1, column 1) and A.sub.peak2,n (UV2, column 2), with n as cycle number. The baselines for the peak areas were drawn from the elution peak start to the UV signal points that corresponded to the end of the elution step in the disconnected state. Since the UV signal was not at zero at this point, the baselines are not horizontal. However, in order to obtain comparable results for A.sub.peak it is more important to draw the baseline in a consistent manner for each peak in each cycle rather than drawing a horizontal baseline.

[0182] The determined areas are listed in Table 1.

[0183] In the ideal case of absolutely identical columns and detectors A.sub.IC1U,n (table 4, column 1) would be equal to A.sub.IC2U,n (table 4, column 3) and A.sub.peak1,n (table 4, column 2) would be equal to A.sub.peak2,n (table 3, column 4). In practice both columns and detectors are significantly different, thus in most cases the areas A.sub.IC1U and Aim are significantly different and the areas A.sub.peak1 and A.sub.peak2 are significantly different from each other. This is confirmed by the ratio A.sub.IC2U/A.sub.IC1U (table 4, column 5) calculated for every single cycle. The ratio shows that A.sub.IC2U in cyclic steady state (cycles 3 and 4) is about 1.3 times larger (30%) than A.sub.IC1U. In the case of different areas it is very important to determine if the difference is due to a detector with different properties, e.g. amplification, or if it is due to a column with different, potentially deteriorated capacity. Making a judgment based on the different areas alone may lead to the erroneous replacement of a column that actually had an acceptable capacity.

[0184] By comparison of the ratios A.sub.peak2/A.sub.peak1 (table 4, column 6) it becomes clear that also A.sub.peak2 is about 1.3 times larger (30%) than A.sub.peak1. Also the ratio of the sums of the breakthrough and the peak areas (A.sub.IC2U,n+A.sub.peak2,n)/(A.sub.IC1U,n+A.sub.peak1,n) is about 1.3 (table 3, column 10).

[0185] Together, this information shows that the detector signal of UV2 is in general 1.3-fold larger than the detector signal of UV1.

[0186] In case the ratios A.sub.IC2U/A.sub.IC1U and A.sub.peak2/A.sub.peak1 would have been different, it would have indicated that the columns differ in capacity. In that case the ratio of the sums of the breakthrough and the peak areas (table 3, column 10) would have been used to determine the difference of the detector signals.

TABLE-US-00004 TABLE 4 Areas and ratios derived from FIG. 5, given in [mAU min]. The columns 5-10 marked with ratio represent ratios of values listed in other columns. For instance the ratio 3/1 (column 5) stands for a division of the values of column 3 by the values of column 1 (thus A.sub.peak2/A.sub.peak1) 1 2 3 4 5 6 7 8 9 10 measured measured measured measured ratio ratio ratio ratio ratio ratio cycle A.sub.IC1U A.sub.peak1 A.sub.IC2U A.sub.peak2 3/1 4/2 2/1 4/3 7/8 (3 + 4)/(1 + 2) 2 128 1157 189 1500 1.48 1.30 9.0 7.9 1.14 1.31 3 111 1112 149 1419 1.34 1.28 10.0 9.5 1.05 1.28 4 113 1112 146 1419 1.29 1.28 9.8 9.7 1.01 1.28

Example 4: Column Capacity Monitoring, Low Impurity Signals

[0187] The twin column countercurrent sequential loading process of example 3 was evaluated for column capacity changes. The feed was not changed during the run and the feed concentration was constant. Thus only the column capacity was potentially changing, for instance due to fouling or harsh cleaning. The explanations refer to the nomenclature used in FIG. 5.

[0188] The fact that A.sub.ICiU,n=A.sub.ICiU,n-1 and/or A.sub.peaki,n=A.sub.peaki,n-1 for both columns (see example 2) indicates the column capacity has stayed constant from one cycle to the other.

[0189] If A.sub.ICiU,n>A.sub.ICiU,n-1 and/or A.sub.peaki,n<A.sub.peaki,n-1 would have been measured with significant difference it would have been indicative of a decreasing column capacity. Typically the accuracy of the area determination is 1%, thus an increase of A.sub.IC1U,n by 1% over A.sub.ICiU,n-1 or a decrease of A.sub.peaki,n by 1% over A.sub.peaki,n-1 cannot be attributed to a loss of capacity. The method may be also used to evaluate the column performance of cycles that are not successive.

Example 5: Column Capacity Monitoring, High Impurity Signals

[0190] A twin column countercurrent sequential loading process was operated for the capture of a product from feed material with a large impurity content. The feed material was the same throughout the entire run (constant product concentration in feed). The operating parameters are summarized in table 5.

TABLE-US-00005 TABLE 5 Operating parameters for twin-column countercurrent sequential loading process of example 5. Q.sub.feed indicates the feed flow rate, Q.sub.buffer indicates the flow rate of all other steps; t is the duration of the substeps of the disconnected phase B and the interconnected phase IC, respectively. The parameters for the final elution correspond to the B parameters except for the feed flow rate, which is zero during the final elution step. The UV detection wavelength was 305 nm. Q.sub.buffer Q.sub.feed Phase step [mL/ buffer [mL/ t [] [] min] [] min] [min] B wash 1.0 25 mM Phosphate, pH 7.0 0.2 6.0 elute 1.0 25 mM Citrate, pH 3.0 0.2 9.0 clean 1.0 0.1M NaOH 0.2 7.0 equilibrate 1.0 25 mM Citrate, pH 3.0 0.2 2.0 1 equilibrate 1.0 25 mM Phosphate, pH 7.0 0.2 4.0 2 IC feed 0 1.0 24.5 wash 1.0 25 mM Phosphate, pH 7.0 0.0 1.0 startup feed 0 1.0 45.0

[0191] Due to the large impurity content of the feed material the detectors were in the non-linear range of detection (over-saturated), as shown in FIG. 6, and it was not possible to determine the areas corresponding to the product break-through in any of the phases of the process (A.sub.ICiU, A.sub.Bi).

[0192] However, based only on the peak areas A.sub.peaki,n it was possible to determine if the column capacity had deteriorated from one cycle to the other. The product elution peak areas for the 2nd and the 3rd cycle of the process were determined to be A.sub.peak2,2=756 mAU min, A.sub.peak2,3=764 mAU min, A.sub.peak1,2=805 mAU min, A.sub.peak1,3=809 mAU min. By calculating the ratios A.sub.peak2,3/A.sub.peak2, 3 and A.sub.peak1,3/A.sub.peak1,2, respectively, it becomes obvious that the areas are identical with a difference of 1% which corresponds to the accuracy of the area determination. This means that the column capacities are identical with a difference of maximum 1%. This is expected since typically the capacity does not change dramatically within two subsequent cycles. The method may be also used to evaluate the columns performance of cycles that are not successive.

[0193] It has to be noted that in the case of large impurity signals it is not possible to decide based on only A.sub.peaki data if the capacities of the two columns are different or if the detectors are give different signals for the same product concentrations. However, if A.sub.peaki measurements are carried out with a constant feed concentration it can be tested by changing the load if the load is in the linear range of product adsorption. It is assumed that no dramatic column capacity decrease takes place within A.sub.peaki measurements of two cycles. If the load is in the linear range of product adsorption and the comparison is made between at least two peak areas A.sub.peaki from each detector the detectors signal difference can be determined by forming the average of the ratios of A.sub.peak2,n/A.sub.peak1,n. The detectors signal difference determination becomes more accurate if the comparison A.sub.peak2,n/A.sub.peak1,n is made for more cycles.

[0194] More information on the linear range of product adsorption is provided in example 10.

Example 6: Feed Concentration and Column Capacity Monitoring, Low Impurity Signals

[0195] The twin column countercurrent sequential loading process of example 3 was evaluated for the effect of product concentration changes in the feed material and simultaneous column capacity changes.

[0196] The operating parameters of the chromatographic process (such as feed flow rate and switch times) were not changed over the cycles but the feed concentration is assumed to be variable and the column capacity is assumed to be constant or to decrease. The different scenarios for possible column degradation and potential feed concentration changes are summarized in table 5.

[0197] In example 3 the breakthrough areas are the same A.sub.ICiU,n=A.sub.ICiU,n-1 and the peak areas are the same A.sub.peaki,n=A.sub.peaki,n-1 for both columns in two consecutive cycles (3 and 4), indicating that neither the feed quality nor the column performance have significantly decreased from one cycle to the other.

[0198] In the following, more examples of relative behavior of A.sub.ICiU and A.sub.peaki and the possible causes with respect to feed concentration and column capacity are provided.

[0199] If A.sub.ICiU,n<A.sub.ICiU,n-1 and A.sub.peaki,n<A.sub.peaki,n-1 is detected the feed concentration must have decreased. The column capacity may have stayed the same from one cycle to the other but may have also decreased. An increasing ratio A.sub.peaki,n/A.sub.ICiU,n from one cycle to the other is indicative of the column quality staying constant or that the capacity is decreasing to a lesser extent than the column load. A decreasing ratio A.sub.peaki,n/A.sub.ICiU,n from one cycle to the other is indicative of the column quality staying constant or the capacity decreasing to a stronger extent than the column load.

[0200] If A.sub.ICiU,n>A.sub.ICiU,n-1 and A.sub.peaki,nA.sub.peaki,n-1 are measured the feed concentration must have increased from one cycle to the other. The column capacity may have stayed the same from one cycle to the other but may have also decreased. An increasing ratio A.sub.peaki,n/A.sub.ICiU,n from one cycle to the other is indicative of the column quality staying constant or that the capacity is decreasing to a lesser extent than the column load increases. A decreasing ratio A.sub.peaki,n/A.sub.ICiU,n from one cycle to the other is indicative of the column quality staying constant or the capacity decreasing to a stronger extent than the column load.

[0201] If A.sub.ICiU,n>A.sub.ICiU,n-1 and A.sub.peaki,n<A.sub.peaki,n-1 are measured the column capacity must have decreased. The feed concentration must have changed such that the load was increased to a stronger extent than the capacity of the columns decreased. Theoretically, the capacity of the columns may decrease to the same extent like the increase of the load such that the effects cancel out. In that case it is recommended to calculate the ratio of the areas corresponding to the total product that is eluting for two different cycles, (A.sub.peaki,n+A.sub.ICiU,n)/(A.sub.peaki,n-1+A.sub.ICiU,n-1), for the same detector. If the ratio is >1, the feed concentration must have increased from one cycle to the other.

TABLE-US-00006 TABLE 5 Changes in column capacity and feed concentration from cycle n-1 to n and the effects on the areas A.sub.ICiU and A.sub.peaki that are determined from the chromatograms. Column capacity changes include: capacity decrease , constant capacity =; Product concentration changes in the feed include the following: increase , decrease in , no change =. columns feed A.sub.ICiU A.sub.peaki = = A.sub.ICiU,n = A.sub.ICiU,n-1 A.sub.peaki,n = A.sub.peaki,n-1 =/ A.sub.ICiU,n < A.sub.ICiU,n-1 A.sub.peaki,n < A.sub.peaki,n-1 = A.sub.ICiU,n > A.sub.ICiU,n-1 A.sub.peaki,n A.sub.peaki,n-1 = A.sub.ICiU,n > A.sub.ICiU,n-1 A.sub.peaki,n < A.sub.peaki,n-1 A.sub.ICiU,n > A.sub.ICiU,n-1 effects may cancel out, check (A.sub.peaki,n + A.sub.ICiU,n)/ (A.sub.peaki,n-1 + A.sub.ICiU,n-1)

Example 7: Feed Concentration and Column Capacity Monitoring, High Impurity Signals

[0202] The twin column countercurrent sequential loading process of example 5 was evaluated for the effect of product concentration changes in the feed material and simultaneous column capacity changes. In the following, the product concentration in the feed will abbreviated with feed concentration

[0203] The operating parameters of the chromatographic process (such as feed flow rate and switch times) were not changed over the cycles but the feed concentration was assumed to be variable and the column capacity is assumed to be constant or to decrease.

[0204] Due to the large impurity content of the feed material the detectors were in the non-linear range of detection (over-saturated), as shown in FIG. 6, and it was not possible to determine the areas corresponding to the product break-through in any of the phases of the process (A.sub.ICiU, A.sub.Bi).

[0205] The fact that A.sub.peaki,n=A.sub.peaki,n-1 for both columns indicates that a.) neither the feed concentration nor the column performance have decreased significantly or b.) that the column capacity for both or one of the columns has decreased to the same extent as the feed concentration has increased from cycle n1 to cycle n.

[0206] If A.sub.peaki,n>A.sub.peaki,n-1 is detected for both or one of the columns indicates that the feed concentration has increased from cycle n1 to n and the column capacity has stayed equal or decreased to a lesser extent as the feed concentration has increased, for both or one of the columns.

[0207] If A.sub.peaki,n<A.sub.peaki,n-1 is detected for both or one of the columns, either the feed concentration has decreased from cycle n1 to n or the column capacity has decreased or both the feed concentration and the column capacity have decreased. Possibly also the feed concentration has increased from cycle n1 to n but the column capacity has decreased to a stronger extent.

Example 8: Process Control, Low Impurity Signals

[0208] Two twin column countercurrent sequential loading processes for the capture of a product from feed material with a large impurity content were evaluated for the effect of product concentration changes in the feed material and simultaneous column capacity changes. In the following, the product concentration in the feed is abbreviated with feed concentration. The twin-column countercurrent sequential loading process conditions are listed in table 7.

TABLE-US-00007 TABLE 7 Operating parameters for the twin-column countercurrent sequential loading processes of example 8. Q.sub.feed indicates the feed flow rate, Q.sub.buffer indicates the flow rate of all other steps; t is the duration of the substeps of the disconnected phase B and the interconnected phase IC, respectively. Q.sub.feed was 0.82 mL/min and 0.19 mL/min, respectively. The parameters for the final elution correspond to the B parameters except for the feed flow rate, which is zero during the final elution step. The UV detection wavelength was 300 nm. Q.sub.buffer Q.sub.feed Phase step [mL/ buffer [mL/ t [] [] min] [] min] [min] B wash 1.0 25 mM Phosphate, pH 7.0 0.82/0.19 6.0 elute 1.0 25 mM Citrate, pH 3.0 0.82/0.19 9.0 clean 1.0 0.1M NaOH 0.82/0.19 7.0 equilibrate 1.0 25 mM Citrate, pH 3.0 0.82/0.19 2.0 1 equilibrate 1.0 25 mM Phosphate, pH 7.0 0.82/0.19 4.0 2 IC feed 0 1.0 24.5 wash 1.0 25 mM Phosphate, pH 7.0 0.0 1.0 startup feed 0 1.0 45.0

[0209] Based on the areas A.sub.ICiU, A.sub.Bi and A.sub.peaki, control actions can be derived to maintain certain levels of A.sub.peaki, which correspond to the concentration of the product of interest in the product pool.

[0210] As outlined in example 6, A.sub.ICiU and A.sub.peaki experience certain trends in response to feed concentration and column capacity changes. The aim of process control is to maintain A.sub.peaki within certain defined limits. The possible control actions include a change of the load and are listed in detail above.

[0211] The control actions based on the trends of the areas A.sub.ICiU and A.sub.peaki are summarized in the following: [0212] If the feed concentration remains constant and the column quality remains constant no control action is required. [0213] If the feed concentration decreases and the column capacity remains the same or decreases the load may be increased to increase A.sub.peaki (see constraint reported below). [0214] In case the feed concentration is increasing and the column quality remains the same or decreases to a lesser extent, the load must be reduced to maintain the desired level of A.sub.peaki. [0215] In case the feed concentration stays equal and the column capacity decreases the load may be increased to increase A.sub.peaki (see constraint reported below). [0216] In case the feed concentration is increasing and the column capacity is decreasing to a larger extent, the load may be increased to maintain the desired level of A.sub.peaki, taking into account the constraint explained below.

[0217] The above cases show that the effects of column capacity deterioration and feed concentration change in the periodic countercurrent loading process can be controlled by changing only the load. Summarizing the above control actions, if A.sub.peaki decreases beyond the specified limits the load should be increased (see constraint below) and if A.sub.peaki increases beyond the specified limit the load should be reduced.

[0218] The constraint for load increases is given by the dynamic binding capacity. The more the capacities of the downstream column in the interconnected state and of the column that is loaded in the disconnected state are exceeded, the more product is lost and the lower the final recovery of the process.

[0219] In order to avoid product losses, the breakthrough of product from the column that is loaded during the disconnected state and the breakthrough of product from the downstream column in the interconnected state must be minimized.

[0220] In process chromatograms these constraints imply that the signals from the columns that are loaded during the disconnected state and the signals from the downstream columns in the interconnected state must not exceed the plateau value that corresponds to the level of the impurities which are not adsorbed. This constraint is graphically illustrated in FIG. 7 for chromatograms of the runs with two different loads (run I: 48 g/L, run II: 30 g/L, only the signal of detector UV2 is shown).

[0221] FIG. 7 shows that the signals from the downstream column in the interconnected phase IC1 (UV2-I and UV2-II) reach the plateau value that corresponds to the non-adsorbing impurities (see arrows in FIG. 7 in phase IC1, for both 30 g/L and 48 g/L load). Thus the areas A.sub.IC2D-I and A.sub.IC2D-II confined by the signal curve and the impurity plateau baseline are not larger than zero and no product is lost in either of the runs during the interconnected state from the downstream column.

[0222] In the disconnected phase B2 the area in between the breakthrough curve and the impurity plateau baseline is zero for run II (area not indicated in FIG. 7 since the signal curve and the impurity plateau baseline match exactly). Thus, in run I, no product is eluted into the waste (and lost) during the B phase. In contrast, for run I, the area indicated with A.sub.B2-I is larger than zero and corresponds to product breakthrough in the loading step of the disconnected phase (i.e. product loss).

[0223] In the interconnected phase IC2 the areas in between the signal curve of the upstream column in the interconnected state and the impurity plateau baseline are designated A.sub.IC2U-I and A.sub.IC2U-II, respectively. In phase IC2 the breakthrough from the previously loaded column either starts (30 g/L load, UV2-II) or respectively continues (48 g/L load, UV2-I), however the column is in the upstream position and the product leaving the column is completely adsorbed in the downstream column. Therefore the product that corresponds to the areas A.sub.IC2U-I and A.sub.IC2U-II, recorded at the upstream column outlet in the interconnected phase, remains in the system and is not lost.

[0224] The area values, obtained by integration, that correspond to product losses are listed in Table 8.

[0225] In order to estimate if the product losses were significant in cyclic steady state the areas A.sub.ICiD and A.sub.Bi were compared with the respective elution peak areas A.sub.peaki by calculating the product loss ratios PLR1=(A.sub.B1+A.sub.IC1D)/(A.sub.peak1+A.sub.B1+A.sub.IC1D) for column 1 (signals not shown in FIG. 7) and PLR2=(A.sub.B2+A.sub.IC2D)/(A.sub.peak2+A.sub.B2+A.sub.IC2D) for column 2 using the area values listed in Table 8 for each of the runs I and II. In this case, since A.sub.IC1D=A.sub.IC2D=0 for both runs I and II, the equations can be facilitated to PLR1=A.sub.B1/(A.sub.peak1+A.sub.B1) and PLR2=A.sub.B2/(A.sub.peak2+A.sub.B2). The average product loss ratio was then calculated as PLRavg=(PLR1+PLR2)/2. For the run with the 48 g/L load, it was found that PLR1 was 14.5% and PLR2 was 11.4% and the average ratio PLRavg was about 13%, indicating a significant product loss. The corresponding product yield was Y=100%-PLRavg=87%. In contrast, all product loss ratios for the 30 g/L run (run II) were zero due to A.sub.B1=A.sub.B2=0 and consequently the yield was 100%.

[0226] The yield values were confirmed by offline HPLC protein A analytics to be 90% (run I) and 100% (run II), which is in good agreement with the values determined by the online area evaluation. In the presented case the evaluation was carried out online by manual integration using the Contichrom evaluation software (ChromaCon AG, Zurich, Switzerland) but it can be fully automated using suitable integration algorithms and automatically trigger control actions for the subsequent cycles.

[0227] Consequently, based on the online analysis results, in run I the load would be lowered by means of a suitable control action in order to avoid product losses in future cycles. Possible control actions include but are not limited to a reduction of the feed flow rate in the disconnected state, a reduction of the feed flow rate in the interconnected state and a reduction of the interconnected state duration, as described further above.

[0228] It is worth noting that depending on the individual column capacity and detector calibration, the areas A.sub.B1 and A.sub.B2B may be significantly different. In order to reliably estimate if potential product losses are significant and require control actions, the areas A.sub.B1 and A.sub.B2 are put into perspective with the corresponding peak areas A.sub.peak1 and A.sub.peak2, respectively, and an average product loss ratio should be estimated based on both columns/detectors. The same applies for the areas A.sub.IC1D and A.sub.IC2D. Furthermore it is worth noting that if A.sub.B1 and A.sub.B2 are zero, also A.sub.IC1D and A.sub.IC2D must be zero (see run II case).

TABLE-US-00008 TABLE 6 Data from the cyclic steady state from two operating points of twin-column sequential loading chromatography process. Run I: 48 g/L load, run II: 30 g/L load. The determined areas correspond to the areas indicated in FIG. 7 for UV2. The areas for UV1 were evaluated too but are not shown in FIG. 7 for clarity. Column 2 shows the feed flow rate in the disconnected state of the process. 10 4 7 9 avg 11 12 2 3 UV1 5 6 UV2 8 A.sub.IC2D product Yield Yield 1 Q.sub.feed,IC Load A.sub.peak1 A.sub.B1 A.sub.IC1D A.sub.peak2 A.sub.B2 [mAU * loss ratio (online) (offline) Run [mL/min] [g/L] [mAU * min] [mAU * min] [mAU * min] [mAU * min] [mAU * min] min] [%] [%] [%] I 0.82 48 344 58.3 0 379 49 0 13.0 87.0 90.5 II 0.19 30 253 0.1 0 282 0 0 0.0 100.0 100.0

Example 9: Optimization of a Twin Column Countercurrent Sequential Loading Process, Low Impurity Signals

[0229] In order to optimize process performance it is desirable to maximize the load of the run in order to process more material in the same amount of time using the same columns.

[0230] In the previous examples suitable criteria for the identification of product losses have been explained in detail. The criteria can serve as optimization tool since they define the maximum load of the process. Based on the criteria, the load of the process may be increased just until the point where product losses start to occur.

[0231] A twin column countercurrent sequential loading process was run for the capture of an IgG monoclonal antibody from clarified cell culture harvest using a Protein A affinity stationary phase packed into two columns of 0.5 cm inner diameter and 5.0 cm length in order to demonstrate the optimization procedure.

[0232] The process was operated with the parameters listed in table 9 on Contichrom Lab-10 equipment from ChromaCon AG, Switzerland.

TABLE-US-00009 TABLE 9 Operating parameters for the twin-column countercurrent sequential loading process of example 9. Qfeed indicates the feed flow rate, Q.sub.buffer indicates the flow rate of all other steps; t is the duration of the substeps of the disconnected phase B and the interconnected phase IC, respectively. Q.sub.feed was increased every two cycles. The parameters for the final elution correspond to the B parameters except for the feed flow rate, which is zero during the final elution step. The UV detection wavelength was 300 nm. Phase step Q.sub.buffer buffer Q.sub.feed t [] [] [mL/min] [] [mL/min] [min] B wash 1.0 25 mM Phosphate. pH 7.0 0.0/0.19/0.43/0.62/0.82 6.0 elute 1.0 25 mM Citrate. pH 3.0 0.0/0.19/0.43/0.62/0.82 9.0 clean 1.0 0.1M NaOH 0.0/0.19/0.43/0.62/0.82 7.0 equilibrate 1 1.0 25 mM Citrate. pH 3.0 0.0/0.19/0.43/0.62/0.82 2.0 equilibrate 2 1.0 25 mM Phosphate. pH 7.0 0.0/0.19/0.43/0.62/0.82 4.0 IC feed 0 1.0 24.5 wash 1.0 25 mM Phosphate. pH 7.0 0.0 1.0 startup feed 0 1.0 45.0

[0233] The same feed material was used throughout the run. The feed flow rate in the interconnected state was kept constant throughout the run at 1.0 mL/min.

[0234] The load flow rate in the disconnected phase was increased after every two cycles in steps from 0.0 mL/min to 0.82 mL/min (see Table 10, column 1). The flow rates corresponded to loads of 26 g/L to 48 g/L (see Table 10, column 2). For column 1, the areas A.sub.peak1, A.sub.B1, A.sub.IC1D were determined (see Table 10, columns 3-5) and for column 2, the areas A.sub.peak2, A.sub.B2 and A.sub.IC2D were determined (see Table 10, columns 6-8) as described in example 3 (Note: for 26 g/L load the areas of UV2 were not determined).

[0235] From the determined areas from UV1 and UV2, the average product loss ratios PLRavg and the yields were calculated (see Table 10, columns 9, 10). For comparison, the yield was determined by offline HPLC analysis (see Table 10, column 11). The productivity (see Table 10, column 12) was calculated as the amount of product produced within one cycle divided by the duration of the cycle and the total column volume based on the offline analysis values.

[0236] The buffer consumption (see Table 10, column 13) was calculated as the amount of buffer consumed (in liters) per gram of product purified based on the offline analysis values. Table 10 shows that with increasing load the productivity increases significantly and the buffer consumption decreases. Increasing the load from 26 g/L to 43 g/L, which corresponds to an increase of more than 65%, still allows for product recovery with more than 97% yield. A further increase of the load to 48 g/L leads to significant product losses of at least 10%, resulting in a product yield of less than 90%. In the presented case, losses of more than 5% were considered inacceptable, so of the tested conditions 43 g/L was the optimal load that had a 65% larger productivity than the base case at 26 g/L and an over 65% reduced buffer consumption. Process yield, determined by online and offline analysis and the productivity are shown as a function of the load in FIG. 8.

[0237] Summarizing, the presented example shows how the sequential loading process can be optimized by increasing the load successively and by monitoring and evaluating the areas under the signal curves corresponding to the positions of potential product loss.

[0238] The effect of a load increase control action becomes evident in the following cycle, potentially allowing for a new control action. In the presented example, the feed concentration was constant but the optimization method can also be applied if the feed concentration changes or the column capacity is different among the columns and/or the capacity of the columns is decreasing. The method can be fully automated using suitable integration algorithms and automatically trigger control actions for the subsequent cycles. Preferably the step size for load increase for process optimization is in the range of 5-10 g/L for an IgG capture using Protein A affinity chromatography. Through successive increase of the load the process the operating space corresponding to high yield is explored. As soon as an operating point with inacceptable yield is found, the process conditions should be reverted to a previous operating point that corresponded to 100% yield. If desired, the process optimization can be continued by increasing the load again by smaller increments or a load safety margin can be left by not continuing the optimization.

[0239] Preferably the process optimization is started from a point that corresponds to an intermediate load. However, if only very little process knowledge is present an optimization starting point is defined by operating parameters corresponding to 0 g/L load in the disconnected phase.

TABLE-US-00010 TABLE 10 Optimization procedure data of the twin-column sequential loading chromatography process (example 9). Column 1 shows the feed flow rate in the disconnected state of the process; column 2 the load; columns 3-5 the areas A.sub.peak1, A.sub.B1, A.sub.IC1D, respectively, of UV1 determined in the 2nd cycle after the feed flow rate change; columns 6-8 the areas A.sub.peak2, A.sub.B2, A.sub.IC2D, respectively, of UV2 determined in the 2nd cycle after the feed flow rate change; column 9 the average product loss ration PLRavg, column 10 the yield determined online by the area evaluation method described in the example, column 11 the yield determined through HPLC offline analysis, column 12 the productivity and column 13 the buffer consumption based on the offline analyses. 3 4 5 6 7 1 2 UV1 UV1 UV1 UV2 UV2 Q Load A.sub.peak1 A.sub.B1 A.sub.IC1D A.sub.peak2 A.sub.B2 [mL/min] [g/L] [mAU * min] [mAU * min] [mAU * min] [mAU * min] [mAU * min] 0.00 26 219 0 0 n.a. n.a. 0.19 30 253 0.1 0 282 0 0.43 37 298 1.7 0 326 2 0.62 43 330 10.5 0 359 10 0.82 48 344 58.3 0 379 49 8 9 10 11 1 UV2 avg product Yield Yield 12 13 Q A.sub.IC2D loss ratio (online) (offline) Prod. B.C. [mL/min] [mAU * min] [%] [%] [%] [g/L/h] [L/g] 0.00 0 0.0 100.0 100.0 14.4 1.2 0.19 0 0.0 100.0 100.0 17.2 1.0 0.43 0 0.6 99.4 99.5 20.7 0.6 0.62 0 2.8 97.2 99.6 23.9 0.4 0.82 0 13.0 87.0 90.5 24.5 0.2

Example 10: Process Control, High Impurity Signals

[0240] The example describes the control of the twin column countercurrent sequential loading process for the capture of a product from feed material with a large impurity signal (see FIG. 6) and variable concentration of product of interest throughout the entire run.

[0241] As shown in example 5, A.sub.B1 and A.sub.IC1D as well and A.sub.B2 and A.sub.IC2D cannot be measured due to the large impurity signal.

[0242] However in the frequently observed case of an only slowly decreasing column capacity, and with implemented safety margins for the load it can be assumed that no product is lost during in the twin-column periodic countercurrent process. Safety margins can be determined by means of offline analysis for product concentration determination. Thus, with reference to example 9, A.sub.B1 and A.sub.IC1D as well and A.sub.B2 and A.sub.IC2D can be assumed to be zero.

[0243] In the described case the area A.sub.peaki changes only in dependence of the load, for instance due to a change of the feed concentration.

[0244] Consequently, based on only the peak areas A.sub.peaki and the comparison of their development over at least two cycles, the process performance can be monitored and control actions can be derived to maintain certain levels of A.sub.peaki, which correspond to the desired concentration of the product of interest in the product pool.

[0245] In case the feed concentration remains constant the peak areas A.sub.peaki remain constant over the cycles (A.sub.peak1,n=A.sub.peak1,n-1 and A.sub.peak2,n=A.sub.peak2,n-1) and no control action is required.

[0246] In case the feed concentration increases, the peak areas A.sub.peaki increase over the cycles (A.sub.peak1,n>A.sub.peak1,n-1 and A.sub.peak2,n>A.sub.peak2,n-1), and a decrease of the load is required if the peak areas have run out of specification.

[0247] In case the feed concentration decreases, the peak areas A.sub.peaki decrease over the cycles (A.sub.peak1,n<A.sub.peak1,n-1 and A.sub.peak2,n<A.sub.peak2,n-1), and an increase of the load is required if the peak areas have run out of specification.

[0248] The control actions may be also based on comparing areas of non-successive cycles or fractions of cycles.

Example 11: Optimization of a Twin Column Countercurrent Sequential Loading Process, High Impurity Signals

[0249] The example describes the optimization of a twin column countercurrent sequential loading process for the capture of a product from feed material with a large impurity signal (see FIG. 6) and a constant concentration of product of interest throughout the entire run. The optimization method is described based on the experimental data shown in example 9 without using the values of the areas A.sub.B1, A.sub.IC1D and A.sub.B2, A.sub.IC2D in order to simulate the presence of a large impurity signal as described in example 5.

[0250] However in order to optimize process performance it is desirable to maximize the load in order to process more material in the same amount of time using the same columns.

[0251] A successive increase of the load inevitably leads to product losses when the column capacity is exceeded. Due to the large impurity signal product losses cannot be directly identified by the areas A.sub.B1, A.sub.IC1D and A.sub.B2, A.sub.IC2D, respectively.

[0252] Nevertheless, process optimization based only on the peak areas A.sub.peaki is possible, when evaluating the peak areas of operating points with different loads.

[0253] The starting point of the process optimization corresponded to a load where no breakthrough was present, namely to a load during the interconnected state that corresponds to a value that is well below the breakthrough value of two sequential columns and with a feed flow rate of zero during the disconnected state. The starting load value corresponded to 26 g/L (see table 11, column 2). As described in example 9 the load was now increased every two cycles and the areas A.sub.peaki were determined by integrating the chromatograms (table 11, column 3 (UV1) and column 6 (UV2)). After having recorded the peak areas A.sub.peak1 corresponding to three loads (26 g/L, 30 g/L and 37 g/L) and two loads (A.sub.peak2, 30 g/L, 37 g/L, the peak area from the first load was not measured), respectively, the area values were plotted as a function of the load and the data was fitted with two straight lines (see FIG. 9). The fact that the peak areas A.sub.peak1 could be fitted with a straight line indicates that starting at 26 g/L additional product loaded led to a proportional increase of the peak areas A.sub.peaki, or in other words that all additional product that is loaded is also adsorbed and that the product losses are negligible. In the presented case, the linear correlation of A.sub.peak1 from 26 g/L to 37 g/L justified also the use of a linear correlation of A.sub.peak2, which was determined using only two points (30 g/L, 37 g/L). The obtained linear correlations between peak areas and load were used to extrapolate the expected peak areas for higher loads (table 11 column 4 (UV1) and column 7 (UV2)). Further points at increased load were recorded and the determined peak areas were compared to the expected peak areas calculated by means of the correlation. At 43 g/L and 48 g/L load the measured area was smaller than the area predicted by means of the correlation. The occurrence of a difference is expected at higher loads since the stationary capacity is limited, which is not taken into account by the linear correlation. The area difference (UV delta) between the expected and measured peak areas is shown in table 11, column 5 (UV1) and column 8 (UV2). The average product loss ratio (PLR) can now be calculated by dividing UV delta by the corresponding expected peak area for each UV and by forming the average of the values of UV1 and UV2. The yield (online, determined by the presented optimization method, see table 11 column 10) can be calculated by 100%PLR or by taking the average of the ratio of measured and expected peak area of the two detectors. For comparison, the yields determined using offline HPLC analyses are also shown (see FIG. 9). A comparison of the two yields (online, offline) shows that the chosen optimization method based on evaluation of only the peak areas was capable of identifying loads that corresponded to negligible product losses and such loads that corresponded to larger product losses. Based on both the presented optimization method and offline analyses the operating point with 48 g/L load would have been ruled out due to inacceptable product losses (>5%) and the operating point corresponding to 43 g/L would have been chosen. As described in example 9, the 43 g/L load corresponded to a 65% larger productivity than the base case at 26 g/L load and to an over 65% reduced buffer consumption.

[0254] It has to be noted that it is not possible to decide based on only A.sub.peaki data if the capacity of the two columns is different or if the detectors have a different signal magnitude. However, if A.sub.peaki measurements are carried out with a constant feed concentration in the linear region of the measurement range and the comparison is made between more than two peak areas A.sub.peaki from each detector, as outlined in this method, information on both detector calibration and column capacity decrease can be obtained.

[0255] Summarizing, the presented example shows how the sequential loading process can be optimized by changing the load and monitoring and evaluating only the peak areas without the need to evaluate areas under breakthrough curves. For this method information of successive cycles is required to determine if the column capacity has been exceeded. The advantage of the method is that it can also be applied in cases where the impurity signal is exceeding the measurement range of the detector. It is recommended to dynamically verify the performance data corresponding to historical operating points if the process is operated over time spans where significant column degradation is expected. For this purpose the load may be changed temporary.

[0256] Both presented optimization methods (example 9 and example 11) do not require detector calibration.

TABLE-US-00011 TABLE 11 Optimization procedure data of the twin-column sequential loading chromatography process (example 11). Column 1 shows the feed flow rate in the disconnected state of the process; column 2 the load; column 3 the area A.sub.peaki determined by integration of the chromatogram of UV1; column 4 the area A.sub.peaki as estimated from the linear correlation described in the example and in FIG. 9 for UV1; column 5 the difference between the two areas of column 4 and column 3 of UV1; column 6 the area A.sub.peaki determined by integration of the chromatogram of UV2; column 7 the area A.sub.peaki as estimated from the linear correlation described in the example and in FIG. 9 for UV2; column 8 the difference between the two areas of column 7 and column 6 of UV2; column 9 the average product loss ration PLRavg, column 10 the yield determined online by the area evaluation method described in the example, column 11 the yield determined through HPLC offline analysis, column 12 the productivity and column 13 the buffer consumption based on the offline analyses. 3 4 6 7 1 2 UV1 UV1 expctd 5 UV2 UV2 expctd Q Load A.sub.peak1 A.sub.peak1 UV1 delta A.sub.peak2 A.sub.peak2 [mL/min] [g/L] [mAU * min] [mAU * min] [mAU * min] [mAU * min] [mAU * min] 0.00 26 219 219 1 n.a. 251 0.19 30 253 252 1 282 282 0.43 37 298 299 0 326 326 0.62 43 330 338 8 359 362 0.82 48 344 377 33 379 399 9 10 11 1 8 avg product Yield Yield 12 13 Q UV2 delta loss ratio (online) (offline) Prod. B.C. [mL/min] [mAU * min] [%] [%] [%] [g/L/h] [L/g] 0.00 n.a. 0.3 99.7 100.0 14.4 1.2 0.19 0 0.2 100.0 100.0 17.2 1.0 0.43 0 0.1 99.0 99.5 20.7 0.6 0.62 3 1.5 98.5 99.6 23.9 0.4 0.82 20 6.9 93.1 90.5 24.5 0.2

TABLE-US-00012 LIST OF REFERENCE SIGNS B Disconnected state or Q.sub.feed,BT1L feed flow rate for single phase (batch state) of the column operation that is sequential loading significantly lower than process maximum desired feed IC Interconnected state or flow rate phase of the sequential EV.sub.1H,1 elution volume loading process corresponding to low t.sub.B duration of the dis- breakthrough based on connected phase of the a single column sequential loading breakthrough curve process recorded for Q.sub.feed,IC t.sub.IC duration of the inter- EV.sub.1L,1 elution volume connected phase of the corresponding to low sequential loading breakthrough based on a process single column break- t.sub.startup duration of the startup through curve recorded phase of the for Q.sub.feed,BT1L sequential loading EV.sub.1H,X elution volume process corresponding t.sub.wash,IC duration of the wash to a high breakthrough phase of the sequential based on a single loading process column breakthrough Q.sub.feed feed flow rale, general curve recorded for Q.sub.feed,IC Q.sub.feed,IC second flow rate, EV.sub.2 elution volume feed flow rate during corresponding the interconnected to low breakthrough phase of the sequential based on a sequentially loading process interconnected column Q.sub.wash,IC wash flow rate during breakthrough curve the interconnected recorded for Q.sub.feed,IC, phase of the sequential multiplied with the loading process safety factor Z. Q.sub.feed,B first flow rate, feed flow EV.sub.Y smaller of the two values rate during the dis- E.sub.1H,X, EV.sub.2 connected phase of the X magnitude of break- sequential loading process through. typically Q.sub.feed, feed flow rate during the 30-90%. .sub.startup startup phase of the W ratio between Q.sub.feed,BT1L sequential loading and Q.sub.feed,IC; Typically process 50-90% typically 60-90% Z safety factor for loading, V.sub.dead dead volume, volume of A.sub.Bi area confined by- interstitial liquid in one breakthrough curve and a column horizontal baseline V.sub.col empty column volume corresponding to plateau c.sub.feed feed concentration signal value of the non- DBC.sub.1H single column adsorbing impurities (EV) breakthrough curve, measured at the outlet of recorded at Q.sub.feed,IC, column i by detector i as a function of the during the disconnected elution volume phase B DB.sub.C2 breakthrough curve of two A.sub.ICiU area confined by (EV) interconnected columns, breakthrough curve and a recorded at Q.sub.feed,IC, as a horizontal baseline function of the elution corresponding to plateau volume signal value of the non- dEV elution volume increment adsorbing impurities for integration measured at the outlet of TL Target load value in the column i by disconnected phase detector i during the of the sequential interconnected phase IC loading process when column i is in the PL Preload value in the upstream position. interconnected phase of the A.sub.ICiD area confined by- sequential loading process breakthrough curve and a A.sub.peak area confined by the elution horizontal baseline peak curve and a horizontal corresponding to plateau baseline corresponding to signal value of the non- the value of the equilibrated adsorbing impurities empty column or a baseline measured at the outlet of defined by other criteria column i by detector such as points of operating i during the parameter changes interconnected phase IC i column index, detector when column i is in the index downstream position. n cycle number PLR product loss ratio