Patent Application
20200054964
The present invention relates to a method for determining operational status, e.g. binding capacities, of a chromatography column using UV detectors with selectable UV cell pathlength.
Binding capacity of a chromatography column for the solute is a very important factor in process chromatography. The binding capacity directly influences the productivity and cost of chromatography step. The binding capacity is defined either in terms of dynamic/breakthrough capacity or as the maximum binding capacity. The dynamic capacity depends on the conditions at which the solution flows through the column packed with chromatography medium, such as residence time defined as the ratio between column volume and feed flow rate. The maximum binding capacity represents a breakthrough capacity of the column if the residence time was at equilibrium. The initial breakthrough capacity is defined as the amount of binding solutes taken up by a column at the point when the solutes are first detected in the effluent. The breakthrough capacity can also be defined as a capacity at a given percentage of breakthrough, where the percentage represents the amount of binding solute present in the effluent from the column expressed in percent of the solute present in the feed. According to this definition the maximum binding capacity will be equal to breakthrough capacity at 100% of breakthrough, i.e., at the point where no more solute can bind to the column. Therefore, in order to determine maximum capacity, the breakthrough capacities are measured at different levels of breakthrough, where the levels are defined by levels of concentration of solutes measured in the effluent from the column during sample loading. Often these concentrations are determined by continuously monitoring a signal in a flowthrough detector placed in the effluent line. The plot of these concentrations (signal) against time (or volume or mass loaded) is called a breakthrough curve. Location of the breakthrough on a chromatogram and its shape is related to how much solute can bind on the column and how quickly all adsorption sites are saturated with the solute. It also shows how much more solute can be bound to the column at any given time. Breakthrough binding capacity for the solute in the presence of the impurities is one of the most critical parameters to optimize when developing a purification protocol. Because impurities more than often have similar light adsorbing properties as the solute determination of binding breakthrough capacities is a tedious and laborious work. In a typical experiment effluent from the column is collected in series of fraction, which are subsequently analyzed for the solute using high resolution techniques, such HPLC. Thus the determination of binding capacities for a chromatography column is rather complicated and in cases where the feed solution concentration is randomly varying during the feed application onto a chromatography column the true breakthrough capacities are intractable or extremely difficult to measure. The determination of binding capacity is very important if one wants to operate a column at the optimum process conditions. For instance, it can be shown that under certain conditions a maximum productivity of a capture chromatography step is obtained when the solute of interest reaches a certain value of its concentration in the column effluent, for instance a 10% of its initial concentration. If the breakthrough capacity is determined according to the method described above, it is impossible to terminate loading of the column at exact 10% breakthrough if either feed concentration or process conditions, including flow rate and/or chromatography media properties, vary with time in unpredictable manner.
Furthermore, determination of breakthrough capacities at different levels of breakthrough under varying process conditions is very important in the case of continuous chromatography.
In continuous chromatography, several identical columns are connected in an arrangement that allows columns to be operated in series and/or in parallel, depending on the method requirements. Thus, all columns can be run in principle simultaneously, but slightly shifted in method steps. The procedure can be repeated, so that each column is loaded, eluted, and regenerated several times in the process. Compared to ‘conventional’ chromatography, wherein a single chromatography cycle is based on several consecutive steps, such as loading, wash, elution and regeneration, in continuous chromatography based on multiple identical columns all these steps occur simultaneously but on different columns each. Continuous chromatography operation results in a better utilization of chromatography resin, reduced processing time and reduced buffer requirements, all of which benefits process economy. Continuous chromatography is sometimes denoted simulated moving bed (SMB) chromatography. Simulated moving bed chromatography is an example of periodic counter current process, because periodically all the chromatography columns comprising the system are simultaneously moved in the direction opposite to the sample flow. The apparent movement of the columns is realized by appropriate redirections of inlet and outlet stream to/from the columns.
Bishop et al (“Simulated Moving Bed technology in Biopharmaceutical Processing”, Bischops,
M. and Pennings, M., Recovery Biological Products XI, (2003) Banff, Alberta, Canada) discloses a continuous chromatography method based on simulated moving bed (SMB) technology, which has been successfully employed for the laboratory scale purification of IgG with a protein A affinity resin. Despite the fact that the multi-column and multi-zone continuous approach provided by SMB greatly increases process efficiency, SMB systems have not been utilized to date for cGMP biopharmaceutical production, mainly because of system complexity from both hardware and operational perspectives. The operational perspective is of particular interest as the continuous methods are more complex and require many operations (steps) to occur simultaneously at very precisely predefined points in time. In contrast to batch chromatography, implementation of safety factors to account for process variability is not recommended for continuous processes as by definitions they operate on the assumption of a steady state that can only be established if there is no variability in the input to the system.
In fact, simulated moving bed technology has been utilised for decades in various other fields. For example, U.S. Pat. No. 3,291,726 (Universal Oil Products) described as early as 1966 a continuous simulated counter-current sorption process for the petrochemical industry.
Heeter et al (Heeter, G. A. and Liapis, A. I., J. Chrom A, 711 (1995)) have suggested, as an alternative to a typical four zone SMB system, a method based on a three column periodic counter-current chromatography (3C-PCC) principle. More recently, Lacki et al (“Protein A Counter-Current Chromatography for Continuous Antibody Purification”, Lacki, K. M. and Bryntesson, L. M., ACS (2004) Anaheim, Calif. USA) described the use of such a 3C-PCC system for IgG adsorption to MabSelect™ affinity resin. This 3C-PCC method requires simpler hardware and easier operation than the typical four zone SMB system, directly reducing the cost associated with the capital equipment and the maintenance of the system.
Historically, essential factors for a reliable continuous process are: 1) the quality of the columns used, and more specifically the similarity or even identity between columns, 2) constant feed composition, and 3) hardware reliance, more specifically constant flow rate delivered by pumps. If the columns are not identical, the theoretical calculations typically used to design continuous chromatography process will not be correct, and it will become difficult to design an efficient and robust continuous chromatography process. The same argument applies if feed concentration and flow rates vary with time in an unexpected manner.
Therefore, for scale-up considerations, having identical columns, reliable pumps in the system is essential. However, the packing of a column with a chromatography media is very complex in order to obtain repeatable results. Even small differences in the number of plates or other packing properties can have a huge effect on the end result. Furthermore, since capacities of chromatography resins typically change during resins lifetime/usage the process conditions chosen for a fresh resin will not be applicable for a resin that has been used several times. If also the feed solution concentration will vary it will be even more complicated to design an efficient continuous chromatography process that would operate at its optimum all the time.
Normally, the composition of the feed material (sample) will vary over time based on the process for producing the sample. Thus, a detector used to determine the composition of the feed material into the chromatography process has to be able to monitor variations in feed composition, i.e. variations in the concentration of target compound in the sample.
An object of the invention is to provide a reliable and dynamic method for determining and monitoring operational status of a chromatography column.
This is achieved by a method according to claim 1. Hereby the UV absorbance of the feed signal and the effluent signal can be used to continuously determine operational status of the chromatography column.
A further object of the invention is to provide a reliable and dynamic method for controlling a chromatography system.
This is achieved in a method according to claim 28. Hereby the chromatography system can be controlled dynamically based on real time measurements of the operational status of the chromatography column.
In one embodiment of the invention the chromatography system is a periodic counter current system.
Further suitable embodiments of the invention are defined in the dependent claims.
a, b, and c shows schematically three valves of
To circumvent the difficulties discussed in relation to prior art a real time control algorithm based on feedback like control principle is provided by the present invention. Accordingly, methods for assessing state of different columns at any given moment of the process are of particular interest. For instance, knowing of an operational status, e.g. binding capacity, of a chromatography column at a particular level of breakthrough would allow one to assess if the column can still bind solutes and how much solute still can be bound before the column reaches full saturation. Similarly, it is of paramount interest from the process yield perspective to know if the initial breakthrough capacity has been reached, as at this point the solute will be found in the effluent stream from the column and, if no proper action is taken, would be direct to waste or would be collected together with other non-binding components.
Furthermore, according to the invention, the first and second detectors 11, 13 are both connected to a determining unit 15. Said unit analyzes the signals detected in the first and second detectors 11, 13 in order to determine the operational status of the chromatography column. Possible signals from the first and second detectors 11, 13 are shown in
The effluent signal is denoted 23 and is the signal from the second detector 13. The effluent signal 23 will start rise from zero at point a, as soon as some of the sample has passed the column 1 and entered the passage of the effluent line 9 where the second detector 13 is positioned. The signal will then rise until point b, where it levels out into a plateau 25. This plateau 25 arises when all the non-binding components in the feed have passed the column. A breakthrough point c is further defined after the plateau 25 when the signal 23 starts to rise again. This is due to the fact that the chromatography media in the column 1 starts to get saturated and some of the parts of the sample that should have been bound in the column start to break through the column. A breakthrough point d is further defined as the signal 23 approaches the signal 21. This point is defined as a saturation point and represents the moment when chromatography media is almost fully saturated with the binding components of the sample.
According to the invention at any given point in time a Delta signal is calculated which is defined to be the feed signal 21 chosen from signals measured between the given time reduced by a specified time delay owing to the specific hardware configuration and the given time minus the effluent signal 23 measured at the given time. The feed signal 21 measures the feature (in one embodiment UV absorbance) for both non-binding and binding components of the feed. The time delay is defined as a time for a non-binding compound in the sample to travel from the feed detector 11 (
According to the invention a Deltasignalmax 27 is calculated which is defined to be the feed signal 21 minus the signal level for the effluent signal 23 when it is in the plateau 25. This Deltasignalmax 27 can then be used for defining suitable levels for the breakthrough point and the saturation point for example. The breakthrough point c can suitably be defined to be a certain predefined percentage of the Deltasignalmax, for example somewhere in the span of 1-10% or more suitably in the span of 1-3% and the Saturation point d can suitably be defined to be a certain predefined percentage of the Deltasignalmax, for example somewhere in the span of 60-90% or more suitable in the span of 70-80%.
One advantage with this approach of determining the breakthrough point and the saturation point is that this could be done automatically in real time and it is independent of the feed concentration and/or composition since the feed signal is automatically compensated for.
In a further aspect of the invention these determinations of operational status, e.g. binding capacities, such as breakthrough and saturation points, are used for automatically controlling the start and stop of the different chromatography process steps, i.e. when a certain breakthrough or saturation point level has been reached a control system can control the chromatography system to proceed to the next process step such as redirecting column effluent to a different collection point, or to stop loading step and initiate column wash step.
In another aspect of the invention the chromatography system comprises more than one chromatography columns, in a so called periodic counter current (PCC) system. In the periodic counter current system, most of the time the feed is passed through at least two columns connected in series. The series is often called a loading zone and addition and removal of columns in/from the loading zone is based on predetermined breakthrough and saturation points for the last and the first column in series, respectively. In
In
In one embodiment of the invention the chromatography process carried out in the system of
(a) Continuously monitoring a signal in the feed line with the first detector 33 and in effluent from each of the columns 39, 47, 59 (with the second, third and fourth detectors 45, 53, 65) and calculating a difference between the feed signal shifted back in time by a predefined delay time and the effluent signals in the outflow line from each column 39, 47, 59;
(b) passing feed comprising at least one target compound across a 1st adsorbent (chromatography media in the first column 39), and directing the outflow from the 1st adsorbent to a 2nd adsorbent (chromatography media in the second column 47) when the Deltasignal (definition as described above in relation to
(c) redirecting the feed to the 2nd adsorbent, and passing wash liquid across the 1st adsorbent to which target compound has bound when the Deltasignal measured between the feed line and the effluent from a 1st adsorbent reaches a predetermined value x2;
(d) directing the wash liquid outflow to the 3rd adsorbent (chromatography media in the third column 59) and subsequently directing the outflow from the 2nd adsorbent to the 3rd adsorbent when the Deltasignal measured between the feed line and the effluent from a 2nd adsorbent reaches a predetermined value x1;
(e) regenerating the 1st adsorbent;
(f) redirecting the feed to said 3rd adsorbent, and passing wash liquid across the 2nd adsorbent to which target compound has bound when the Deltasignal measured between the feed line and the effluent from a 2nd adsorbent reaches a predetermined value x2;
(g) directing the wash liquid outflow to the 1st adsorbent, and subsequently directing the outflow from the 3rd adsorbent to the 1st adsorbent when the Deltasignal measured between the feed line and the effluent from a 3rd adsorbent reaches a predetermined value x1;
(h) regenerating the 2nd adsorbent;
(i) redirecting the feed to said 1st adsorbent, and passing wash liquid across the 3rd adsorbent to which target compound has bound when the Deltasignal measured between the feed line and the effluent from a 3rd adsorbent reaches a predetermined value x2;
(j) directing the wash liquid outflow to the 2nd adsorbent, and subsequently directing the outflow from the 1st adsorbent to the 2nd adsorbent when the Deltasignal measured between the feed line and the effluent from a 1st adsorbent reaches a predetermined value x1;
(k) regenerating the 3rd adsorbent;
(l) repeating steps (b)-(k);
wherein at least one target compound is collected in step (d), (g) and/or (j).
Predetermined values of x1 and x2 represent breakthrough and saturation points, respectively.
The current invention enable use of not identical columns when operating a counter current system because any differences in the columns properties can be compensated for by automatically adjusting breakthrough and saturation switching points based on the Deltasignal and Deltsignalmax measured for each of the columns. It also enables operating a counter current system when unexpected changes in feed concentration occur as any change in the feed concentration, and thus a change in the mass loaded into each column can be compensated for by automatically adjusting the breakthrough and saturation switching points based on Deltasignal and Deltasignalmax that automatically compensates for variation in feed concentration.
In another embodiment of the invention, the chromatography system comprising of more than 2 chromatography columns can be used for direct capture of a product from a feed stream originated from a perfusion cell culture. For a person skilled in art, it is well known that concentrations of components in such stream will vary with time, and without an automated control algorithm operation of the chromatography system would be impossible without a risk of significant losses of product due to wrongly a priori assigned redirection points.
The present example is provided for illustrative purposes only, and should not be construed as limiting the present invention as defined by the appended claims.
This example illustrates a continuous primary capture step for purification of a monoclonal antibody (MAb) from a mixture containing MAb and bovine serum albumin, BSA, on protein A chromatography resin using a four column periodic counter current (4C-PCC) system with deltaUVmax control according to the invention (i.e. in this example the detectors are UV detectors and the Deltasignalmax is called deltaUVmax). More specifically, four similar columns were packed with the Protein A chromatography resin MabSelect™ (GE Healthcare Bio-Sciences, Uppsala, Sweden). The columns were connected to a custom modified AKTAexplorer™ (GE Healthcare Bio-Sciences, Uppsala, Sweden) chromatography system (
The following single column chromatography cycle was used as a base for operating the 4C-PCC system in a continuous manner: 1) column equilibration with 3 column volume (CV) of buffer A; 2) column loading with feed; 3) column wash with 4CV of buffer A; 4) column elution with 4CV of buffer B; 5) column CIP with 4CV of buffer C; and, 6) column regeneration with 3 CV of buffer A. All steps were performed at 0.4 mL/min flow rate.
Composition of solutions used is given below:
A few hundred milliliters of solution containing the feed were continuously fed into the experimental 4C-PCC setup described above. The absorbance of the feed solution was measured continuously by the first UV detector 115 positioned on the feed line (
The repetitive UV pattern shown in
Summary of results obtained in this experiment is shown in Tables 1 and 2 below. Where the volume of feed solution and the calculated amount loaded onto each column is shown. These data are compared to the calculated results that would be obtained if no automated controlled according to this invention was used, i.e., using predesignated switching times of flow redirection as described in point 1-18 were used if the reference run were performed on the same system with the same columns. As shown in
Because the first column 107 was different than the second column 109 a premature breakthrough for this column would have been observed in an experiment when the 4C-PCC system was operated based on predefined column switching times. Under such conditions, a loss in MAb during step 1 of the cycle would have been observed. In addition, the amount of MAb loaded onto the second column 109 in step 2 would be larger than calculated and subsequently some of the MAb would be lost in the effluent from the second column 109 before this column would be connected to the third column 111 in step 4. This mismatch in amounts loaded onto the columns would then propagate through steps 5-18, and the amount of MAb not capture by the columns would steadily increase with each column switch. This loss in unbound MAb was avoided by implementing the control algorithm according to the current invention. In Table 1, estimated masses of MAb loaded onto each of the columns during the experiment are shown. The masses loaded were estimated based on the areas above respective UV curves measured in the effluent lines after each column. The mass washed out from one column and loaded on the second to the next column in series was neglected. As shown in Table 1 significantly different mass of MAb was loaded on to the first column 107 as compared to the mass loaded on the other three columns. The mass loaded on the first column 107 was between 20-30% smaller than the mass loaded on each of the other columns during two different cycles that was almost the same with no more than 5% difference between columns and cycles. If on the other hand, the 4C-PCC system was operated without the automated control with a switch time set to enable load of 77 mg per milliliter resin, the amount of MAb lost per cycle and column would be significant (Table 2) and would account for around 10% of mass loaded on to the system during the whole experiment.
A standard ultraviolet (UV) detector for liquid chromatography measures the absorbance of monochromatic light of fixed wavelength in the UV or visible wavelength range (typically between 190 nm and 400 nm) against a reference beam and relates the magnitude of the absorbance to the concentration of a compound in the sample passing through a flow cell contained within the detector.
Compounds suitable for UV detection typically contain unsaturated bonds, aromatic groups, or functional groups containing heteroatoms, which contain π* and σ* nonbonding orbitals into which electrons are promoted to absorb the incident energy. These nonbonding orbitals contain a wide distribution of vibrational and rotational energy levels that lead to a distribution of absorbance energies and therefore spectra with broad, rather than sharp features.
Compound concentration can be determined from the Beer-Lambert law as described in more detail below.
The detection optics is comprised of an input light guide 240 with a light exit surface 250 arranged adjacent and in optical alignment with a light entrance 260 surface of an output light guide 270. As is disclosed, the input light guide 240 and the output light guide 270 protrudes into the flow cell channel 230 so that the optical gap between the light exit surface 250 and the light entrance surface 260 is situated centrally in the flow cell channel 230. The optical gap may e.g. be situated at any position in the flow cell channel 230, as long as it is not located in a stagnant zone, e.g. adjacent a wall of the channel in order to ensure that the fluid in the gap is representative of the fluid flow. In the disclosed embodiment, the input light guide 240 and the output light guide 270 protrudes essentially transverse into the flow cell channel 230, whereby exchange of fluid in the optical gap is promoted. However, the angular relationship between the fluid flow channel and the light guides 240 and 270 need not be transverse, but can be any suitable angle and configuration provided that desired renewal of the fluid in the optical gap is achieved. In alternative embodiments, the fluid flow channel 230 need not be straight, but it may be curved or the like.
It has surprisingly been found that the present approach with relatively small sized light guides 240 and 270 protruding into the fluid flow channel 230 provides excellent capability of measuring high concentrations of e.g. proteins in the flow when arranged at a short distance from each other such as 0.5 mm or less. As will be discussed in more detail below, a high degree of linearity for high concentrations has been disclosed for embodiments with an optical gap of 0.12 and 0.07 mm. The optical gap is selected depending on the concentration range to be measured, and it may e.g. be any value from 0.02 to 2 mm such as 0.7, 0.5, 0.4, 0.3, 0.2, 0.1, 0.07, 0.05 mm or there between. It has been found that the fluid in the optical gap is readily exchanged even at the smallest gaps and at moderate flow rates, which is believed to be a result of the small optical cross sectional area and the “protruded” position of the gap essentially at the center of the fluid flow channel.
In
As is mentioned, it is believed to be essential to avoid disruption of the fluid flow, whereby a small cross sectional area of the protruding portions of the light guides and the output light guide in the flow direction is essential, and in the disclosed embodiments the total cross sectional area of the protruding portions of the input and output light guides 240, 270 is less than about A/2, or even A/4 in the flow direction.
The light guides may be comprised of any suitable optical material capable of transmitting light, such as glass, quarts, light transmitting polymers etc. One or both may be provided with an optically blocking sleeve or coating (not shown) at the peripheral surface to avoid leakage of light. The light guides 240 and 270 may be cylindrical with a diameter of 5 mm or less, e.g. 3 mm, or 1 mm or the like. The diameter of the light guides 240 and 270 obviously depends on the total area A of the flow channel 230. The cross sectional flow area of the fluid flow channel, schematically disclosed in
As can be seen in
In
UV cell pathlength of the detector defines an operating window having an upper limit, and wherein the step of adapting the UV cell pathlength further comprises reducing the UV cell pathlength while maintaining the estimated maximal concentration of target compound below the upper limit.
The concept behind the invention is to provide a method for determining operational status, e.g. binding capacities, of a chromatography column, comprising:
The feed signal is generated using a first UV detector having a first UV cell pathlength operating at a first UV wavelength and the effluent signal is generated using a second UV detector having a second UV cell pathlength operating at a second UV wavelength. The method further comprises.
The advantage of the above mentioned features is that the possibility to measure the composition of the feed material even when the composition of feed material varies over time, or is higher than originally anticipated.
The step of determining the first threshold value may comprise:
The feed material comprises impurities and product, and the absorbance measured by the first UV detector is a combination of the contribution from both. The impurities results in background absorbance when measured in the first UV detector, and the product results in an absorbance level of the product in the feed material when measured in the first UV detector.
In one aspect the absorbance level of the product is at least forty percent (40%) of the background absorbance of the feed material. In another aspect, the absorbance level of the product in the feed material is higher than the background absorbance, preferably the absorbance level of the product in the feed material is 70%-90% of the absorbance measured in the first UV detector.
The step of selecting the first UV cell pathlength may comprise selecting a fixed UV cell pathlength from a plurality of pre-selected UV cell pathlengths, or adjusting the UV cell pathlength. However, the step of selecting the first UV wavelength may also comprise adjusting the first UV wavelength.
According to one aspect, the step of selecting the first UV cell pathlength and/or first UV wavelength is performed until a suitable operating window of the first UV detector is achieved. This is achieved when the first UV detector is within a linear range of detection.
In a first embodiment, the UV absorbance of the feed material is calculated, and the step of estimating maximum UV absorbance is based on the calculated UV absorbance. The calculation may be based on the Beer-Lambert Law (see above).
In a second embodiment the UV absorbance of the feed material is measured, and the step of estimating maximum UV absorbance is based on the measured UV absorbance. The measurement may be performed before running a batch in the chromatography system to calibrate the UV detector. Alternatively, the UV absorbance of the feed material is continuously measured, and the first threshold value is continuously selected based on changes in UV absorbance.
The step of selecting the first threshold value may be performed when the change in UV absorbance is more than a pre-determined percentage compared to the first threshold value. As an example, the pre-determined percentage may be 1%, 2%, 5%, 10%, 15% or 20% of the threshold value used to select the UV cell pathlength. In a particular embodiment, the UV detector comprises a plurality of pre-selected UV cell pathlengths. This may be realised by a UV detector provided with several parallel light guides having different UV cell pathlengths. In another embodiment, the distance between the light guides may be adjusted to obtain different UV cell pathlengths in response to desired characteristics.
Thus, if the UV detector measure an increase in UV absorbance of a pre-determined percentage, e.g. 5%, then the UV cell length is reduced to ensure that the composition of the feed material still can be detected.
The method further comprising detecting the feed signal and the effluent signal using the same type of detector, i.e. UV detectors, to detect the UV absorbance in the effluent. The effluent signal is generated using a second UV detector having a second UV cell pathlength operating at a second wavelength, the method further comprising:
Preferably, the second threshold value is lower or equal to the first threshold value.
According to one embodiment, the step of determining the second threshold value comprises:
The breakthrough point is an indication that the chromatography column is full and the feed of material into the column has to stop or be shifted to another column.
The UV absorbance for the selected breakthrough point may be calculated, and the step of estimating UV absorbance for the selected breakthrough point in the effluent is based on the calculated UV absorbance for the selected breakthrough point.
In a further embodiment the method further comprises measuring the UV absorbance in the effluent, preferably continuously, and the step of estimating UV absorbance for the selected breakthrough point is based on the measured UV absorbance.
In an embodiment, the method further comprises selecting the second UV cell pathlength to be equal to the first UV cell pathlength.
Similar to the first UV detector, the step of selecting the second UV cell pathlength may comprise selecting a fixed UV cell pathlength from a plurality of pre-selected UV cell pathlengths, or adjusting the UV cell pathlength. However, the step of selecting the second UV wavelength may also comprise adjusting the second UV wavelength.
According to one aspect, the step of selecting the second UV cell pathlength and/or second UV wavelength is performed until a suitable operating window of the second UV detector is achieved. This is achieved when the second UV detector is within a linear range of detection.
The method may further comprise, when the first UV cell pathlength and/or first UV wavelength is changed, applying a time delay corresponding to a residence time in the column before the second UV cell pathlength and/or second UV wavelength is changed.
Advances in cell culture technology have enabled the production of antibody titers upwards of 30 g/L. These highly productive cell culture systems can potentially lead to productivity bottlenecks in downstream purification due to lower column loadings, especially in the primary capture chromatography step. Alternative chromatography solutions to help remedy this bottleneck include the utilization of continuous processing systems such as periodic counter current chromatography (PCC).
Recent studies have provided methods to optimize and improve the design of PCC for cell culture titers up to 5 g/L. This disclosure defines an experimental strategy for purification of a range of cell culture titers up to about 41 g/L. Initial experimentation showed an inability to determine a difference in change in UV280 nm signal (i.e. ΔUV) between cell culture feed and monoclonal antibody (mAb) concentration for column loadings up to at least 100 g/L. Further investigation revealed UV280 absorbance of the cell culture feedstock without antibody was outside of the linear range of detection. Additional experimentation showed the difference in ΔUV for various cell culture feedstocks can be either theoretically predicted by Beer's Law given a known absorbance of the media background and impurities or experimentally determined using various UV280 cell pathlengths. Based on these results, a 0.35 mm pathlength was chosen for dynamic control. Dynamic binding experiments showed a difference in experimental versus theoretical antibody breakthrough curves that was heavily dependent on the feed type and shape of breakthrough chromatograph. Based on the DBC and AUV data, the breakthrough challenge for column experiments was set at ΔUV of 70%. Performance metrics, yield, and purity, were acceptable between various cell culture feeds and titers up to about 31 g/L, which showed ΔUV can dynamically control and define loading in a continuous chromatography process. At titers above 31 g/L, the linearity declines for a 0.35 mm UV pathlength at a UV280 nm signal but retains linearity at a UV300 nm signal using the same pathlength. Therefore, the UV300 nm signal could be a detection option for controlling loading of feeds greater than 31 g/L. The strategy and results presented in this disclosure show column loading in a continuous chromatography step can be dynamically controlled independent of the cell culture feedstock and titer and allow for enhanced process control for implementation into downstream continuous operations.
Transition from batch to continuous processing has proven very successful in many industries, yet its implementation into biologics manufacturing has currently been relatively deliberate [1-3]. This can be attributed mainly to a perception that continuous technology is still in its infancy (i.e. an emerging technology) and due to uncertainties regarding regulatory demands for moving from batch production to continuous production [4, 5]. Additional concerns for adaptation are centered around an increased complexity in operation and process control [6].
The trend towards continuous processing, as seen in many other industries, is clear from an economic perspective since it can achieve higher productivities at shorter processing times with reduced total costs, reduced equipment footprint, and consistent product quality [7]. Continuous processing has been extensively studied and established for fermentation operations using perfusion culture technology but limited examples exist for downstream continuous processing [8-10]. For example, recent literature has demonstrated a continuous process from a perfusion bioreactor through to the downstream polishing steps using periodic counter-current chromatography for the capture and polishing chromatography steps US20140255994(A1), Genzyme Corporation, hereby incorporated by reference. Current approaches for downstream continuous chromatography include simulated moving bed chromatography (SMB), periodic counter-current chromatography (PCC), and sequential multi-column chromatography (SMCC) [11-14]. These enabling technologies refer to a chromatography capture step, utilizing multiple columns which are periodically moved against the direction of feed flow. The main difference of each technology is manifested in the design of the actual hardware between different multicolumn chromatographic systems. One main challenge is that limited information, process controls, and methodologies exist for implementation of these chromatography technologies from laboratory to manufacturing scales to enable a continuous downstream process.
One key facet for operating a continuous chromatography process involves the determination of a steady state protein loading onto each column. Steady state loading allows for minimal protein losses in the process resulting in higher productivities compared to a batch process. Previous solutions for controlling protein loading during continuous chromatography involve either a time based or dynamic control mode. Time base mode involves protein loading as a function of a set value of time, which can result in higher protein losses as a function of resin re-use assuming a decrease in dynamic binding capacity with time [15]. In addition, time base mode for protein loading also does not take into consideration any potential variability in performance amongst columns. Dynamic control of protein loading is based on the relative difference in ultraviolet (UV) signal between the column inlet and outlet UV absorbance (i.e. ΔUV) converted into a percentage. Using this relationship allows loading to a defined level of breakthrough, in order to maximize the column loading to ensure minimal product losses and consistent protein concentrations in each chromatography elution pool [16]. One challenge with dynamic control as cited in the literature is that a minimal difference in UV signal of impurities and antibody can occur for different cell culture feedstocks, which significantly decreases the robustness and utility of dynamic control. In turn, this low signal to noise ratio forces the user to operate in a time based mode for protein loading [17]. Throughout this specification, the term protein is used and should be construed in a general manner to comprise proteins in general and to include but not be limited to antibodies, monoclonal antibodies, antibody conjugates and protein conjugates.
Previous work in the literature has demonstrated the use of multivariable wavelength detectors to control and equalize the antibody breakthrough signal given a perfusion titer of about 2-3 g/L [18]. This allows the user to achieve a consistent delta UV percentage across columns to eliminate variability in column loading [18]. One drawback of this approach for dynamic control is that the root cause of the interference problem would need to be investigated and determined at various wavelengths, which can be an intractable challenge with current cell culture media designs and continuous upstream cell culture optimization. Another challenge of a multivariable wavelength detector is the protein absorbance will be significantly reduced, which may limit the range of feed titer that can be dynamically controlled by the UV detector. Our discovery builds upon the initial principle of UV control by providing a solution for overcoming the difference in UV signal between impurities and proteins, which allows the user to utilize dynamic control for protein loading in a continuous chromatography process. Furthermore, in this disclosure, the ability to dynamically control protein loading for high concentration cell culture feedstocks up to about 31 g/L is also demonstrated, and additional data indicates that dynamic control is possible for titers greater than 31 g/L if a UV300 nm wavelength were utilized.
Investigation of impurity versus antibody absorbance levels for a variety of different UV cell pathlengths were evaluated and compared against theoretical calculations to determine the optimal pathlength for signal detection and control. Continuous chromatograph experiments were executed using different harvest impurity feedstocks and titers to determine the ability to dynamically control column loading independent of different types of harvest feedstocks. These sets of data were utilized to define a continuous chromatography process, which can dynamically control protein loading independent of the cell culture feedstock, generate consistent quality elution pools, and allow for enhanced process control for implementation into downstream continuous operations.
Materials & Methods
Harvested Cell Culture Fluid (HCCF)
The antibodies used in this study along with cell culture media are listed in Table I.
Each fermentation batch consisted of a 14-day fed-batch process with harvest viability of at least 60%. Each batch was harvested by an optional centrifugation step followed by depth filtration and sterile filtration. The material was then stored frozen and thawed prior to use. A portion of the HCCF was ultrafiltered to reach some of the desired feed concentration set points if necessary. The UV background of the impurities was not significantly affected by this concentration step.
Protein A Chromatography Process
The chromatography resin used throughout this study was MabSelect SuRe (GE Healthcare, Uppsala, Sweden), which is an agarose-based Protein A affinity matrix. The columns were equilibrated with 3 column volumes (CVs) of 10 mM sodium phosphate pH 6.5. After loading of harvested cell culture fluid, the columns were washed with equilibration buffer for 3 CVs. To remove product-related impurities, the columns were washed with 10 mM sodium phosphate, 0.5M sodium chloride pH 6.5 for 3 CVs followed by equilibration buffer for 3 CVs. The antibody was eluted using 20 mM sodium acetate pH 3.5 for about 3-4 CVs. The columns were cleaned using a combination of 100 mM acetic acid pH 2.9 and 0.1N NaOH. All columns were stored in 200 mM sodium acetate, 2 v % benzyl alcohol pH 5.5.
Reverse Phase High Performance Liquid Chromatography (RP-HPLC)
The analytical RP-HPLC method consisted of a R2/10, 2.1 mm×30 mm column (Applied Biosystems) with column temperature maintained at 70° C. The antibody concentration was determined using the following gradient at a flowrate of 2 mL/min: Mobile Phase A: 0.2 v % trifluoroacetic acid (TFA) in water, Mobile Phase B: 0.2 v % TFA in 90% acetonitrile, Method: 68% A to 40% A in 5 minutes followed by 1 minutes at 100% B, and then 2 minutes at 68% A. UV280 nm absorbance was recorded during each injection and peaks were integrated using Chemstation software (Agilent Technologies). Antibody concentration was determined using a standard curve.
Ultra Performance Size Exclusion Chromatography (UP-SEC)
The analytical UP-SEC method consisted of a BEH 200, 4.6×150 mm column (Waters) with column temperature maintained at ambient temperature (15-30° C.). Monomer, dimer, and higher order aggregate separation was obtained in 100 mM sodium phosphate, 100 mM sodium chloride pH 7.2 with a flow rate of 0.5 mL/min for 5 minutes. UV214 nm absorbance was recorded during each injection and peaks were integrated using Chemstation software (Agilent Technologies). The monomer purity percentage was determined by the monomer peak divided by the total peak area. The aggregate content percentage was determined by the sum of the peak area of each aggregate peak divided by the total peak area.
High Performance Ion Exchange Chromatography (HP-IEX)
Analytical high performance ion exchange chromatography (HP-IEX) method was performed using a 4 mm×250 mm Dionex cation exchange ProPac WCX-10 column (Waltham, Mass., USA) on an Agilent 1200 HPLC system. The main monomer peak was separated from the acidic and basic mAb variants using a linear gradient with increasing salt concentration and pH. Acidic variants are defined as species with pl less than the average pl of the monomer whereas basic variants are defined as specifies with pl higher than the average pl of the monomer. The main peak percentage was determined using UV280 nm by dividing the main monomer peak area by the total peak area. The acidic or basic variant percentages were determined by the sum of the acidic or basic variant peak areas respectively divided by the total peak area. Acidic and basic variants can be composed of monomers and aggregates.
Ultraviolet (UV) Absorbance Measurements of HCCF
UV280 nm measurements were taken of both the HCCF feeds and samples without antibody (i.e. Protein A flowthrough) ranging from 3 to 41 g/L using pathlengths from 0.05 to 2 mm using the Solo VPE (C Technologies, Bridgewater, N.J.). A curve of absorbance versus pathlength was created to determine linearity of the feeds. The theoretical antibody absorbance was also measured by Beer's Law and compared to the experimental value for various cell culture feeds. Beer's Law is described by the following equation:
A=E
1%
×b×c
where A=absorbance (Au)
and c=protein concentration (g/L)
Dynamic Binding Chromatography (DBC) Experiments
HCCF was loaded onto about a 4 mL column (0.5 cm×20 cm) containing MabSelect SuRe resin up to a resin loading of about 120 grams of protein per liter of resin (g/L) to ensure full antibody breakthrough. Fractions were collected during loading and analyzed using RP-HPLC to determine percentage of antibody breakthrough. Linear velocity ranges were studied from 100 to 600 cm/hr. For the dynamic binding experiments, antibody breakthrough percentage was calculated using the following equation:
where BT=antibody breakthrough (%)
Cloading fraction=antibody concentration by offline RP-HPLC of each fraction during column loading (g/L)
and CHCCF=antibody concentration by offline RP-HPLC of HCCF (g/L)
The dynamic binding capacity was predicted for various antibody feed concentrations and linear velocities with the Method Design Tool (GE Healthcare, Uppsala, Sweden) using the pore diffusion model. Theoretical DBC results were compared against experimental DBC results to determine the accuracy of the pore diffusion model as a function of antibody feed concentration.
Periodic Counter-Current Chromatography (PCC) Experiments
The PCC chromatography system was operated in the 3-column mode using three 4 mL columns.
In the second phases of the closedown block, columns 1 and 2 are washed, eluted, and regenerated separately, as indicated in
Dynamic Control
The dynamic control function utilizes the difference in UV absorbance between the total UV absorbance from the product and the background, and the background of the feed. This difference is defined as AUV.
UV cell was calibrated against the UV Prod cell prior to experimentation. The ΔUV was set to 70% for the initial experiments and then ranged from about 20-45% for subsequent experiments.
Results & Discussion
Evaluation of Dynamic Binding Capacity for Various HCCF Feedstreams
Prior to testing any of the HCCF feeds with the 3-column PCC method, the relationship between antibody breakthrough as a function of column loading and ΔUV was investigated. A dynamic binding capacity (DBC) experiment was conducted using the high UV absorbance HCCF to determine the ΔUV difference of impurities versus antibody.
Since the media impurities had a very high background absorption outside the linear range of the UV detector, the DBC experiment did not show any difference in ΔUV as indicated in
In order to confirm this hypothesis, Beer-Lambert's Law was utilized to calculate the theoretical antibody absorbance for two different types of mAb #1 HCCF feed streams at 0.5 mm versus 2 mm UV cell pathlengths. For the high UV absorbance feed, the theoretical mAb #1 absorbance was 0.85 Au but the HCCF components plus antibody was at least 4.4 AU, which is beyond the linear range and explains why antibody breakthrough could not be detected in the DBC experiment, as indicated in Table II. However, at 280 nm and a pathlength of 0.5 mm, the theoretical antibody absorbance decreased from 0.85 AU to 0.21 AU, which should be sufficient to allow for dynamic control using ΔUV. The HCCF absorbance without antibody decreased significantly within the linear range to 1.6 AU. Another DBC experiment was performed using the 0.5 mm UV280 nm pathlength and resulted in a AUV of about 0.22 AU, which correlated exactly to the increase in antibody breakthrough detected through offline RP-HPLC analysis.
For the mAb #1 low UV absorbance HCCF feed stream, Beer Lambert's law indicates the theoretical antibody absorbance at 0.41 Au will be sufficient for ΔUV control at 2 mm pathlength but the theoretical antibody absorbance at 0.10 Au will be too low for accurate ΔUV control at the 0.5 mm pathlength. A DBC experiment was performed using the 2 mm pathlength cell and confirmed a ΔUV difference of 0.42 Au, which matched the calculated antibody absorbance of 0.41 Au from Beer's law, as indicated in
However, the column loading of the mAb #1 high absorbance feed can only be dynamically controlled using a lower UV cell pathlength, for this study a pathlength of 0.5 mm was sufficient. In order to utilize a lower UV cell pathlength in future PCC experiments, the linearity and ability to sufficiently control the ΔUV was investigated in connection with
Determination of UV Cell Pathlength and Linearity for Dynamic Control of Various HCCF Feeds
A prerequisite for reliable real time monitoring of the UV absorbance during column loading or elution is that the UV absorbance signal is within the linear range. This is not a limitation from running PCC or continuous chromatography. Rather it is an inherent characteristic of the feed used in the process, where both the antibody titer as well as the background of the impurities, media and components will influence the available range for the dynamic control. Since it was demonstrated that Beer's law can be used to predict if ΔUV control will be successful for different HCCF feed streams for a given antibody, the absorbance of different HCCF feedstocks for various antibodies were measured to determine both the ΔUV difference and linearity versus UV cell pathlength. This information will provide valuable insight to determine if the column loading of different HCCF feedstocks at various titers from 3 to 41 g/L can be dynamically controlled in a PCC mode, as indicated in Table I.
The UV280 nm absorbance of each HCCF sample including a sample without antibody (i.e. Protein A flowthrough) was measured for pathlengths ranging from 0.05 to 2 mm. For both high UV absorbing feeds (mAb #1 and mAb #3), the absorbance of the sample without antibody loses linearity at a 1 mm pathlength (as seen in the previous study). As the UV cell pathlength decreases, the absorbance of the blank sample becomes linear as a function of UV cell pathlength but there is a minimal AUV difference at pathlengths from 0.05 to 0.15 mm since the antibody absorbance significantly decreases as indicated in
For the mAb #1 HCCF sample containing the lowest total UV absorption, the AUV difference is more pronounced even at lower pathlengths (down to 0.35 mm) if target protein exceeds 4 g/L as indicated in
The ability to dynamically control protein loading using the ΔUV functionality for a 3C-PCC column system was tested using the four HCCF feeds listed in Table I. In order to challenge the ability of the UV280 nm UV cell to control protein loading, the AUV percentage was set at 70% for all of the runs. Performance metrics for the column experiments included a comparison of yield and purity levels to the control, which is a batch mode process to determine the success of the dynamic control function. Protein loading and UV280 nm difference between the UV pre and post column in the start-up phase and within the loop were monitored for each run to aid in determination of the ability to control protein loading using the UV280nm shorter pathlength cells.
This higher loading in the start-up phase could result in some antibody that is not captured by the second column in the loading zone and hence lost in the flowthrough as indicated in Table IV.
However, either a lower setpoint for ΔUV or possibly setting the baseline value earlier during the loading will decrease the protein loading and should increase the antibody recovery to within a few percentage of the batch mode control levels. The 41 a medium UV absorbance HCCF feed may be the only exception since the UV capabilities of this feed was shown to be outside of the linear range described in connection with
The ability of the UV pre and post sample cells to consistently control the AUV setpoint is shown in
For HCCF titers up to 10 g/L, the column loading is consistent across each column as shown by the ÄKTA chromatograph, see
Additional studies were conducted to determine the impact of the ΔUV setpoint on the process performance of the Protein A chromatography step and quality of the FNVIP pools. The lower range of the ΔUV was set at 20% based on the DBC curves in connection with
Yield loss across the Protein A chromatography step decreased substantially by about 13-19% when the ΔUV increased from 20 to 45% for HCCF titers of 20 g/L, see
The UP-SEC purity of each FNVIP pool was consistent and independent of ΔUV range from 20-70%, see
According to one embodiment, a method of separating a protein from a harvested cell culture fluid (HCCF) is defined. The method comprises feeding the HCCF into a continuous chromatography system comprising at least two columns. The system comprising a first UV detector upstream said columns and a second UV detector downstream the column receiving the HCCF and the system is arranged to compare signals registered by the first and second UV sensors to provide a delta UV response indicative of breakthrough of protein through the column.
Each UV detector has a UV pathlength operating at a UV wavelength, and the system is arranged to switch the feed of HCCF to a different column when the delta UV signal reach a predetermined breakthrough point. The pathlength of the UV detector is in the range 0.05-2.0 mm, the breakthrough point is in the range 5-90% delta UV response when the UV wavelength is in the range 280-300, and the concentration of the protein in the HCCF is up to 41 g/l. The pathlength may in particular be selected to be 0.05, 0.10, 0.15, 0.20, 0.35, 0.50, 1.0, or 2.0 mm
The protein includes but are not limited to monoclonal antibody, protein conjugates and antibody conjugates. In case where the protein is a monoclonal antibody, the method may comprise adjusting, when initial signal of HCCF is greater than 3.0 AU, the UV cell pathlength and/or UV wavelength until a linear signal is obtained. A linear signal may be obtained by adjusting the UV cell pathlength and/or UV wavelength until the UV signal is below about 2.5 AU.
The continuous chromatography system may be selected from the group: periodic counter current chromatography, simulated moving bed chromatography, continuous counter-current tangential chromatography, and sequential multi-column continuous chromatography.
Preferably, the continuous chromatography system is selected to be a periodic counter current chromatography system.
When the background absorbance is less than 2.0 Au and the concentration of the protein in the HCCF is up to 31 g/L, the method may comprise selecting the pathlength to be ≤2.0 mm, and the wavelength to be 280 nm. Preferably, the pathlength may be selected to be between 0.05-0.35 mm.
When the background absorbance is less than 2.1 Au and the concentration of the protein in the HCCF is up to 41 g/L, the method may comprise selecting the pathlength to be ≤1.0 mm, and the wavelength to be 280 nm. Preferably, the pathlength may be selected to be between 0.05-0.20 mm.
It should be noted that the concentration of the protein may be higher than 41 g/L provided the linearity of the UV absorbance is maintained.
When the background absorbance is less than 2.0 Au and the concentration of the protein in the HCCF is up to 26 g/L, the method may comprise selecting the pathlength to be ≤0.5 mm, and the wavelength to be 280 nm. It should be noted that the concentration of the protein may be higher than 26 g/L provided the linearity of the UV absorbance is maintained.
When the background absorbance is less than 1.2Au and the concentration of the protein in the HCCF is up to 41 g/L, the method comprises selecting the pathlength to be 2 mm, and the wavelength to be 300 nm. It should be noted that the concentration of the protein may be higher than 41 g/L provided the linearity of the UV absorbance is maintained.
It should be mentioned that the process of dynamic loading can be applied to fed-batch, intensified, and perfusion antibody cell culture streams. Basically, the dynamic control is feed independent. Furthermore, the claimed methods and systems may also be expanded to include non-antibody modalities and also cover resin modalities, including but not limited to Protein A resin, anion exchange, mixed mode, and HIC. Example of Protein A resins include, but are not limited to, Mabselect, Mabselect Sure, Mabselect Sure PCC, Mabselect Sure LX, Mabselect Xtra, and TOYOPEARL AF-rProtein A HC-650F.
Abbreviations
PCC: Periodic counter-current chromatography
3C-PCC: Three column periodic counter-current chromatography
UV: Ultraviolet
DBC: Dynamic binding capacity
RP-HPLC: Reverse phase high performance liquid chromatography
HCCF: Harvested cell culture fluid
UP-SEC: Ultra performance size exclusion chromatography
Med: Medium
mAb: Monoclonal antibody
HCP: Host cell protein
SMB: Simulated moving bed chromatography
SMCC: Sequential multi-column chromatography
HP-IEX: High performance ion exchange chromatography
FNVIP: Filtered neutralized viral inactivated Protein A product
2. Laird, T., Continuous Processes in Small-Scale Manufacture. Org. Process Res. Dev., 2007. 11(6): p. 927.
3. Masak, J., Ganapathy Subramanian (Ed.): Continuous Processing in Pharmaceutical Manufacturing. Chem. Listy. 109(6): p. 475.
4. Allison, G., Regulatory and Quality Considerations for Continuous Manufacturing. Presentation at the International Symposium on Continuous Manufacturing of Pharmaceuticals: Implementation, Technology, & Regulatory (Cambridge, Mass.), 2014.
5. Chatterjee, S., FDA Perspective on Continuous Manufacturing. Presentation at the IFPAC Annual Meeting (Baltimore, Md.), 2012.
6. Hernandez, R., Continuous Manufacturing: A Changing Processing Paradigm. BioPharm International, 2015. 28(4): p. 20-27, 41.
7. Godawat, R., et al., End-to-end integrated fully continuous production of recombinant monoclonal antibodies. J Biotechnol. 213: p. 13-9.
8. Li, L., et al., A single-use, scalable perfusion bioreactor system. BioProcess Int., 2009. 7(6): p. 46, 48, 50, 52-54.
9. Voisard, D., et al., Potential of cell retention techniques for large-scale high-density perfusion culture of suspended mammalian cells. Biotechnol. Bioeng., 2003. 82(7): p. 751-765.
10. Walther, J., et al. Multi-staged experimental design for efficient process development of a continuous perfusion cell culture: American Chemical Society.
11. Angarita, M., et al., Twin-column CaptureSMB: a novel cyclic process for protein A affinity chromatography. J Chromatogr A. 1389: p. 85-95.
12. Godawat, R., et al., Periodic counter-current chromatography—design and operational considerations for integrated and continuous purification of proteins. Biotechnol. J. 7(12): p. 1496-1508.
13. Lacki, K. M. and L. M. Bryntesson. Protein A periodic counter-current chromatography for continuous antibody purification. 2004: American Chemical Society.
14. Ng, C. K. S., et al., Design of high productivity sequential multi-column chromatography for antibody capture. Food Bioprod. Process. 92(2): p. 233-241.
15. Trilisky, E., et al. Capacity assessment: DBC is not the whole story—implications for resin lifetime and sequential multicolumn chromatography: American Chemical Society.
16. Blom, H., et al. The use of dynamic control in periodic counter current chromatography: American Chemical Society.
17. Pollock, J., et al., Optimising the design and operation of semi-continuous affinity chromatography for clinical and commercial manufacture. J Chromatogr A. 1284: p. 17-27.
18. Warikoo, V., et al., Integrated continuous production of recombinant therapeutic proteins. Biotechnol Bioeng. 109(12): p. 3018-29.
19. Cooper, R. S. and D. A. Liberman, Fixed-bed adsorption kinetics with pore diffusion control. Ind. Eng. Chem., Fundam., 1970. 9(4): p. 620-3.
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2018/053659 | 2/14/2018 | WO | 00 |
| Number | Date | Country | |
|---|---|---|---|
| 62461347 | Feb 2017 | US |
