WO 2006/099308 PCT/US2006/008919 A Method of Weak Partitioning Chromatography Field of the Invention [0001] The invention relates to methods of recovering a purified product from a load fluid including one or more impurities. In certain embodiments of the invention, the methods comprise passing the load fluid through a medium at operating conditions which cause the medium to bind at least 1 mg of product per mL of medium, and recovering the purified product in the column effluent during the load cycle and any essentially isocratic wash. In other embodiments of the invention, the methods comprise passing the load through a medium at operating conditions defined by a partition coefficient of at least 0.1. Background of the Invention [0002] Within the biotechnology industry, the purification of proteins on a commercial scale is an important challenge to the development of recombinant proteins for therapeutic and diagnostic purposes. Problems related to yield, purity, and throughput plague the manufacturing sector. With the advent of recombinant protein technology, a protein of interest can be produced using cultured eukaryotic or prokaryotic host cell lines engineered to express a gene encoding the protein. What results from the host cell culturing process, however, is a mixture of the desired protein along with impurities that are either derived from the protein itself, such as protein variants, or from the host cell, such as host cell proteins. The use of the desired recombinant protein for pharmaceutical applications is contingent on being able to reliably recover adequate levels of the protein from these impurities. [0003] Conventional protein purification methods are designed to separate the protein of interest from impurities based on differences in size, charge, solubility, and degree of hydrophobicity. Such methods include chromatographic methods such as affinity chromatography, ion exchange chromatography, size exclusion chromatography, hydrophobic interaction chromatography, immobilized metal affinity chromatography, and hydroxyapatite chromatography. These methods often employ a separation medium that can be designed to selectively adhere either the protein of interest or the impurities. In the bind- elute mode, the desired protein selectively binds to the separation medium and is differentially eluted from the medium by different solvents. In the flow-through mode, the impurities specifically bind to the separation medium while the protein of interest does not, thus allowing the recovery of the desired protein in the "flow-through." 1 WO 2006/099308 PCT/US2006/008919 [0004] Current methods for the purification of proteins, such as antibodies, include two or more chromatographic steps. For example, the first step in the protein purification protocol often involves an affinity chromatography step that utilizes a specific interaction between the protein of interest and an immobilized capture reagent. Protein A adsorbents are particularly useful for affinity capture of proteins, such as antibodies, which contain an Fc region. However, drawbacks to using Protein A chromatography for protein purification include leakage of the Protein A capture agent, leading to contamination of the eluted protein product. Additionally, affinity capture does not separate protein variants, such as aggregated forms of the protein, from the protein of interest. [0005] Researchers have used bind-elute methods, flow-through methods, and displacement methods in efforts to recover proteins free from impurities resulting from both the culturing process and from possible prior steps in the purification process itself. Examples of groups using a bind-elute step as a typical second step to purifying proteins after an affinity capture step include: US Patent 4,983,722, describing a bind-elute ion exchange method of reducing Protein A from a mixture; US Patent 5,429,746, describing a bind-elute hydrophobic interaction chromatography method for purifying IgG antibody from a mixture including Protein A impurities; and US Patent 5,644,036, describing a three-step process for obtaining a purified IgG antibody preparation comprising a Protein A step, a bind-elute ion exchange step, and a size exclusion step. Other groups have used a flow-through step after the affinity chromatography step. For example, PCT publication WO 04/076485 describes a method for removing leaked Protein A from an antibody purified by a Protein A chromatography step followed by a flow-through ion exchange step. PCT publication WO 03/059935 describes a method for purifying a protein in a sample comprising subjecting the sample to a flow-through hydroxyapatite chromatography step following an affinity chromatography step. [0006] Other groups have used a single polishing-step purification scheme to avoid the problems associated with prior purification steps. For instance, US Patent 6,177,548 describes a single-step flow-through ion exchange method for removing aggregates from a biological sample where the pH of the sample is adjusted to 0.2 logs below the isoelectric point of the biological sample. US Patent 5,451,662 describes a single-step bind-elute ion exchange method where the pH of the crude protein mixture is adjusted to a point between the ranges of isoelectric points of the protein fractions to be separated. PCT publication WO 05/044856 describes a single-step displacement method for removal of high molecular weight aggregates from antibody preparations using hydroxyapatite chromatography. 2 ' WO 2006/099308 PCT/US2006/008919 [0007] None of the conventional bind-elute or flow-through methods in the prior art, however, is able to meet the needs of the biotechnology industry in terms of all the requirements of throughput, yield, and product purity. Bind-elute methods and displacement methods are limited by, among other factors, the capacity limit of the separation medium for the desired protein. Flow-through methods, on the other hand, do allow for higher load challenges than bind-elute methods but are limited by the capacity of the separation medium for the impurities. With flow-through methods, no substantial binding of the product to the column occurs; any substantial product binding is seen as negatively impacting product recovery. There is still a need for methods of recovering purified proteins at high throughput that meet the requirements for purity and yield necessary for therapeutic and diagnostic applications. In addition, commercial manufacturing processes add the needs for reliable, robust, and cost-effective purification schemes. Summary of the Invention [0008] The present invention relates to methods of recovering a purified product from a load fluid including one or more impurities by passing the load fluid through a medium at operating conditions which cause the medium to bind at least 1mg of product per mL of medium and recovering the purified product in the column effluent during the load cycle and any essentially isocratic wash. In other embodiments, the operating conditions cause the medium to bind at least 5 mg of product per mL of medium. In another embodiment, the operating conditions cause the medium to bind at least 10 mg of product per mL of medium. In other embodiments, the operating conditions cause the medium to bind at least 20, 30, 40, 50, or 60 mg of product per mL of medium. [0009] The present invention also relates to methods of recovering a purified product from a load fluid including one or more impurities by passing the load fluid through a medium at operating conditions defined by a partition coefficient of at least 0.1 and recovering the purified product in the column effluent during the load cycle and any essentially isocratic wash. In one embodiment, the partition coefficient is in the range of about 0.2 to about 20.0. In another embodiment, the partition coefficient is in the range of about 0.2 to about 10.0. In another embodiment, the partition coefficient is in the range of about 1.0 to about 5.0. In another embodiment, the partition coefficient is in the range of about 0.5 to about 5.0. In an additional embodiment, the partition coefficient is in the range of about 0.5 to about 1.5. 3 WO 2006/099308 PCT/US2006/008919 [0010] The present invention also relates to methods of recovering a purified product from a load fluid including one or more impurities by passing the load fluid through a medium at operating conditions which cause the medium to bind from at least 1 to about 70 mg of product per mL of medium and defined by a partition coefficient of 0.3 to 20, and recovering the purified product in the column effluent during the load cycle and any essentially isocratic wash. [0011] The invention also provides for identifying, in a screening step, the operating conditions that cause the medium to bind at least 1 mg product per mL of medium or alternatively, are defined by a partition coefficient of at least 0.1. The screening step can employ batch binding studies or column binding studies, such as gradient elution studies or isocratic elution studies. [0012] Operating conditions include pH levels, ionic strengths, salt concentrations, excipient concentrations (such as phosphate concentrations, calcium concentrations, arginine concentrations, glycine concentrations, and HEPES concentrations), and counterligand levels (such as imidazole concentrations), depending on the selection of medium. [0013] The medium can be any type of chromatographic resin or separation medium, including a charged ion exchange medium, such as an anion exchange medium or a cation exchange medium, a hydrophobic interaction chromatography resin, a hydroxyapatite resin, or an immobilized metal affinity chromatography resin. [0014] Purified products that can be recovered using the invention include fusion proteins, Fc-containing proteins, immunoconjugates, cytokines, interleukins, hormones, and therapeutic enzymes. [0015] Impurities that can be removed using the invention include host cell proteins, nucleic acids, product variants, endotoxins, Protein A, and viruses. [0016] In one embodiment, the medium removes at least 99.9% of the impurities in the load fluid including host cell proteins, nucleic acids, product variants, endotoxins, and Protein A. [0017] In another embodiment, the concentration of product variants in the purified product is no more than about 2%. [0018] In additional embodiments, the load onto the medium may be at a load challenge of at least 500 mg or at least 1000 mg of product per mL of medium. [0019] In one aspect of the invention, a purified product is recovered from a load fluid including one or more impurities by passing the load fluid through a charged ion exchange medium at operating conditions comprising pH levels and ionic strengths which 4 WO 2006/099308 PCT/US2006/008919 cause the medium to bind at least 1 mg of product per mL of medium or alternatively, at operating conditions defined by a partition coefficient of at least 0.1. [0020] In another aspect of the invention, a purified product is recovered from a load fluid including one or more impurities by passing the load fluid through a hydrophobic interaction chromatography resin at operating conditions comprising pH levels, ionic strengths, and salt concentrations which cause the medium to bind at least 1 mg of product per mL of medium or alternatively, at operating conditions defined by a partition coefficient of at least 0.1. [0021] In another aspect of the invention, a purified product is recovered from a load fluid including one or more impurities by passing the load fluid through a hydroxyapatite chromatography resin at operating conditions comprising pH levels, ionic strengths, phosphate concentrations, calcium concentrations, arginine concentrations, glycine concentrations, HEPES concentrations, and imidazole concentrations which cause the medium to bind at least 1 mg of product per mL of medium or alternatively, at operating conditions defined by a partition coefficient of at least 0.1. [0022] In yet another aspect of the invention, a purified product is recovered from a load fluid including one or more impurities by passing the load fluid through an immobilized metal affinity chromatography resin at operating conditions comprising counterligand levels and pH levels which cause the medium to bind at least 1 mg of product per mL of medium or alternatively, at operating conditions defined by a partition coefficient of at least 0.1. [0023] The methods of the invention can be optionally combined with one or more purification steps. The optional step(s) can be performed either prior to or following the practice of the inventive method. For example, the methods of the invention can optionally be combined with a Protein A chromatography step as an initial step. [0024] In one embodiment of the invention, a product-containing fluid is eluted from a Protein A column using an elution buffer of low ionic strength; the pH and conductivity of the product-containing fluid is adjusted using a neutralization buffer which results in no more than 20mM of the ionic strength of the product-containing fluid, resulting in the load fluid; and the load fluid is passed through an anion exchange medium under the operating conditions of the invention. [0025] In some embodiments, the elution buffer comprises molecules with a charged cationic group with a pKa of 6.5-10. In other embodiments, the elution buffer further comprises molecules with a charged anionic group with a pKa of 2-5. In certain 5 WO 2006/099308 PCT/US2006/008919 embodiments, the elution buffer comprises molecules which are zwitterions at pHs between 7 and 9. [0026] The invention also provides for purified products, including purified proteins and antibodies, prepared by the methods of the invention. [0027] Additional objects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objects and advantages of the invention will be realized and attained by means of the elements and combinations particularly pointed out in the appended claims. [0028] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed. [0029] The accompanying drawings, which are incorporated in and constitute part of this specification, and together with the description, serve to explain the principles of the invention. Brief Description of the Figures [0030] Figure 1 shows (A) the relationship between a partition coefficient and a product adsorption isotherm; and (B) adsorption isotherms for product binding to resin, for three modes of operation: bind-elute mode, weak partitioning mode, and flow-through mode. [0031] Figure 2 shows (A) the partitioning regions for three modes of operation in ion exchange chromatography: bind-elute mode, weak partitioning mode, and flow-through mode; and (B) the partitioning regions for three modes of operation in hydroxyapatite. [0032] Figure 3 shows schematic chromatograms for three modes of operation: bind- elute mode, weak partitioning mode, and flow-through mode. [0033] Figure 4 shows a comparison between weak partitioning and flow-through chromatograms. [0034] Figure 5 shows (A) typical contaminant removal profiles as a function of Kp; and (B) recovery as a function of load challenge and Kp. [0035] Figure 6 shows typical progression of weak partitioning chromatography step development, including 1) high throughput screen to determine Kp, 2) low load challenge runs, 3) high challenge capacity runs, and 4) optimal weak partitioning chromatography runs. [0036] Figure 7 shows a contour plot of Kp vs. pH and the total chloride concentration from the low concentration dataset, as described in Experiment 1.1. 6 WO 2006/099308 PCT/US2006/008919 [0037] Figure 8 shows Protein A removal as a function of the partition coefficient, Kp, as described in Experiment 1.1. The log removal value increases with Kp. Flow-through mode is indicated by the dashed box with "FT", while weak partitioning mode is indicated by the dashed box with "WP." [0038] Figure 9 shows a contour plot of log10Kp vs. pH and the log of the total chloride concentration, as described in Experiment 2.1. [0039] Figure 10 shows (A) for Mab-AAB, host cell protein breakthrough profiles as a function of Kp in ion exchange chromatography; and (B) for Mab-AAB, Protein A breakthrough as a function of Kp in ion exchange chromatography. [0040] Figure 11 shows for Mab-MYA, the optimum operating window for weak partitioning chromatography in hydroxyapatite. The optimum Kp in this example is between 1.5 and 20. [0041] Figure 12 shows for Mab-A5T4, the optimum operating window for weak partitioning chromatography in hydroxyapatite. The optimum Kp in this example is between 2 and 20. [0042] Figure 13 shows for Mab-MYO, the optimum operating window for weak partitioning chromatography in hydroxyapatite. The optimum Kp in this example is between 5 and 20. Detailed Description of the Invention A. Definitions [0043] In order that the present invention may be more readily understood, certain terms are first defined. Additional definitions are set forth throughout the detailed description. [0044] The term "flow-through mode" refers to a product preparation separation technique in which at least one product contained in the preparation is intended to flow through a chromatographic resin or medium, while at least one potential contaminant or impurity binds to the chromatographic resin or medium. Generally, the product partition coefficient for flow-through mode is less than 0.1 and bound product concentration is < 1 mg/mL. The "flow-through mode" is an isocratic operation. [0045] The term "bind-elute mode" refers to a product preparation separation technique in which at least one product contained in the preparation binds to a chromatographic resin or medium. Generally, the product partition coefficient for bind-elute 7 WO 2006/099308 PCT/US2006/008919 mode is greater than 20 and the bound product concentrations are between 1 - 20 mg/mL. The bound product in this mode is eluted during the elution phase. [0046] The term "weak partitioning mode" refers to a product preparation separation technique in which at least one product contained in the preparation, and at least one contaminant or impurity, both bind to a chromatographic resin or medium. The binding of product in weak partitioning mode is at least 1 mg of product per mL of chromatographic resin or medium. Generally, the product partition coefficient for weak partitioning mode is at least 0.1. The "weak partitioning mode" is an isocratic operation. [0047] The term "partition coefficient" (Kp) refers to the equilibrium ratio of the concentration of product absorbed to the resin (Q) to the concentration of product in the solution (c), under specified conditions of pH and solution composition. The partition coefficient Kp is also related to the product adsorption isotherms as shown in Figure 1. The partition coefficient Kp corresponds to the slope of the product adsorption isotherm at very low solution concentrations. It is related to the maximum capacity as follows: where Qmax is to maximum capacity of the resin for the product, and kd is the dissociation constant for 'resin - product' interaction. The partition coefficient is typically measured with a batch binding technique, but other techniques, such as isocratic chromatography, can be used. [0048] The term "bound product" (Q) refers to the amount of product which binds to the resin when in equilibrium with a feedstream. [0049] The term "antibody" refers to any immunoglobulin or fragment thereof, and encompasses any polypeptide comprising an antigen-binding site. The term includes, but is not limited to, polyclonal, monoclonal, monospecific, polyspecific, non-specific, humanized, human, single-chain, chimeric, synthetic, recombinant, hybrid, mutated, grafted, and in vitro generated antibodies. The term "antibody" also includes antibody fragments such as Fab, F(ab')2, Fv, scFv, Fd, dAb, and other antibody fragments that retain antigen-binding function. Typically, such fragments would comprise an antigen-binding domain. [0050] In certain embodiments of the invention, the antibody is one which comprises a CH2/CH3 region and therefore is amenable to purification by Protein A chromatography. The term "CH2/CH3 region" refers to those amino acid residues in the Fc region of an immunoglobulin molecule which interact with Protein A. In some embodiments, the 8 WO 2006/099308 PCT/US2006/008919 CH2/CH3 region comprises an intact CH2 region followed by an intact CH3 region, and in other embodiments, comprises a Fc region of an immunoglobulin. Examples of CH2/CH region-containing proteins include antibodies, immunoadhesions and fusion proteins comprising a protein of interest fused to, or conjugated with, a CH2/CH3 region. [0051] The term "load" refers to any load material containing the product, either derived from clarified cell culture or fermentation conditioned medium, or a partially purified intermediate derived from a chromatography step. The term "load fluid" refers to a liquid containing the load material, for passing through a medium under the operating conditions of the invention. [0052] The term "impurity" refers to any foreign or objectionable molecule, including a biological macromolecule such as a DNA, an RNA, or a protein, other than the protein of interest being purified that is also present in a sample of the protein of interest being purified. Impurities include, for example, protein variants, such as aggregated proteins, high molecular weight species, low molecular weight species and fragments, and deamidated species; other proteins from host cells that secrete the protein being purified (host cell proteins); proteins that are part of an absorbent used for affinity chromatography that may leach into a sample during prior purification steps, such as Protein A; endotoxins; and viruses. [0053] The term "essentially isocratic wash" refers to a solution which varies only slightly from the load fluid in terms of composition or pH. [0054] The term "column effluent" refers to the liquid exiting the medium or column during the load cycle, or in the period that the load is being applied. [0055] The term "load challenge" refers to the total mass of product loaded onto the column in the load cycle of a chromatography step or applied to the resin in batch binding, measured in units of mass of product per unit volume of resin. [0056] The term "log removal value" (LRV) refers to the log(base 10) of the ratio of the mass of impurity in the load of a purification step to the mass of impurity in the product pool. [0057] The term "isocratic chromatography" refers to the operation of a chromatographic column with a solvent that does not change strength during the period of interest. 9 WO 2006/099308 PCT/US2006/008919 B. Description of the Method [0058] The present invention provides methods for recovering purified products from a load fluid containing one or more impurities. The invention has application to the large- scale preparation of proteins for therapeutic and diagnostic purposes. 1. Weak partitioning mode [0059] Applicants have surprisingly found that by operating in a chromatographic mode residing in the region between conventional bind-elute and flow-through chromatography modes, a high degree of impurity reduction, as well as high product load challenge and product recovery, can be obtained. Applicants have named this intermediate product binding mode, the "weak partitioning mode." [0060] In weak partitioning mode, a load fluid containing a product of interest and one or more impurities is passed through a chromatographic medium, with both the product and the impurities binding to the medium. However, the impurities bind more tightly to the medium than the product and as loading continues, unbound product passes through the medium and is recovered from the column effluent. The medium is optionally subsequently washed under isocratic conditions to recover additional weakly bound product from the medium and the purified product from any essentially isocratic wash is pooled with the purified product from the column effluent during the load cycle. [0061] In accordance with the invention, weak partitioning mode is defined by operating conditions which cause the medium to bind at least 1 mg of product per mL of medium. In one embodiment, the operating conditions cause the medium to bind at least 5 mg of product per mL of medium. In another embodiment, the operating conditions cause the medium to bind at least 10 mg of product per mL of medium. In another embodiment, the operating conditions cause the medium to bind at least 20 mg of product per mL of medium. [0062] In certain embodiments of the invention, the total product mass bound to the medium is at least 10% of the total product mass loaded onto the medium. In some embodiments, the total product mass bound to the medium is at least 20% of the total product mass loaded onto the medium. In other embodiments, the total product mass bound to the medium is at least 30% of the total product mass loaded onto the medium. [0063] In accordance with the invention, weak partitioning mode is also defined by a partition coefficient of at least 0.1. In some embodiments, operating in weak partitioning mode comprises operating under conditions defined by a partition coefficient in the range of about 0.2 to about 20.0. In certain embodiments, operating in weak partitioning mode comprises operating under conditions defined by a partition coefficient in the range of about 10 WO 2006/099308 PCT/US2006/008919 0.2 to about 10.0. In other embodiments, operating in weak partitioning mode comprises operating under conditions defined by a partition coefficient in the range of about 1.0 to about 5.0. In other embodiments, operating in weak partitioning mode comprises operating under conditions defined by a partition coefficient in the range of about 0.5 to about 5.0. In yet other embodiments, operating in weak partitioning mode comprises operating under conditions defined by a partition coefficient in the range of about 0.5 to about 1.5. [0064] At least one embodiment of the present invention provides weak partitioning mode operating conditions which cause the medium to bind from at least 1 to about 70 mg of product per mL of medium, and which are defined by a partition coefficient of 0.3 to 20. [0065] Figure 1 shows the product adsorption isotherms for the bind-elute, flow- through, and weak partitioning modes, with product binding for weak partitioning mode being clearly intermediate in comparison to bind-elute and flow-through modes. Because the value of the product partition coefficient (Kp) is the ratio of the concentration of the adsorbed product to the concentration of the product in solution, the Kp values for weak partitioning mode are also intermediate to the values for bind-elute and flow-through modes. [0066] Figure 2A depicts the partitioning regions for bind-elute, weak partitioning, and flow-through modes as a function of ionic strength, showing that Kimp is higher in weak partitioning mode than in flow-through mode. Under the more stringent binding conditions of weak partitioning mode, a higher product capacity can be achieved - higher than flow- through mode, as impurities are more strongly bound, and higher than bind-elute mode, as the product binds very weakly in comparison to impurities and does not take up the majority of the resin capacity. The impurity partition coefficient (Kimp) is higher at more stringent binding conditions, resulting in lower concentrations of residual impurities in the product pool of weak partitioning mode compared to the product pool of flow-through mode. The flow-through, weak partitioning and bind-elute regions in hydroxyapatite, as a function of phosphate and NaC1 concentration, are shown in Figure 2B. [0067] Table A summarizes the differences in characteristics between the three modes of binding: bind-elute (B-E), weak partitioning (WP), and flow-through (FT). 11 WO 2006/099308 PCT/US2006/008919 Table A Characteristics of FT/WP/B-E modes FT WP B-E Kp <0.1 0.1 - 20 >20 Load challenge limitation Impurities 10-50 mg Prod/mL (typical) but actually dependent on load purity Impurities 50- 500 mg Prod/mL (typical) but actually dependent on load purity Product + impurities < 100 mg Prod/mL Load Vol Moderate, for dilute impurities' 10-20 CVs Very high, for dilute impurities up to 50 CVs Lower, as the product binds in addition to impurities 5-20 CVs [Product] in load eluate Equal to load concentration through much of load Initial lag, then equal to load concentration through much of load <5% of load concentration Residual [Impurity] Low Very low Dependent on elution conditions, pool volume and capacity. Product bound (Q) < 1 mg/mL < 10 - 20 mg/mL > 10-20 mg/mL Operating region Relatively broad range of conditions Modest window of operation between FT and B-E modes Stringent binding conditions for load, broad range of elution conditions Mobile phase(s) Isocratic Isocratic Change in buffer composition after load which causes elution. [0068] Weak partitioning mode can also be distinguished from bind-elute and flow- through modes by their chromatograms, as shown in Figure 3. At first, the chromatograms for flow-through and weak partitioning modes may seem quite similar - the product is recovered in the column effluent and wash fractions, under isocratic conditions. However, subtle, but meaningful distinctions exist in the chromatograms which can be used to distinguish these modes, as shown in Figure 4. There is a delay in the initial breakthrough profile (> 0.1 column volumes or CV) for weak partitioning mode compared to flow-through mode. There is a slower washout profile in weak partitioning mode. A small strip peak containing product may be present (which corresponds to the resin still binding 10 - 50% of the load product concentration after the wash stage), which can be recovered from the resin by applying a 1 - 5 CV wash after the load under isocratic conditions subsequent to recovery of the column effluent during the load cycle. [0069] Figure 5A shows the general trends in contaminant LRV for various levels of product partition coefficient values. Contaminant LRVs are relatively low at Kp conditions corresponding to flow-through operations. Operating under conditions of increasing Kp significantly increases LRVs in the column effluent fractions prior to contaminant breakthrough. As shown in the examples, operating at higher Kp values improves the 12 WO 2006/099308 PCT/US2006/008919 contaminant LRVs by as much as 2 logs from those corresponding to the standard flow- through conditions. [0070] Increasing Kp typically increases both the product as well as contaminant binding to the resin. The stronger binding of the contaminant at higher Kp leads to a greater LRV in the column effluent fractions prior to contaminant breakthrough. However, the load challenge at the point of contaminant breakthrough decreases with increasing Kp as the product begins to compete with the contaminant for the binding sites on the resin, as schematically represented in Figure 5 A by the high Kp curve. The weak partitioning region therefore corresponds to an operating window that balances the improvement in contaminant LRV with column capacity requirements for a given separation. [0071] The upper Kp limit for weak partitioning chromatography is also dependent on the column load challenge as shown in Figure 5B. The partition coefficient has no impact on product recovery at values bordering flow-through conditions. The product recovery begins to drop at high Kp values where the isocratic wash conditions are is not effective at washing the bound product off the column in a reasonable number of wash volumes. The extent of product loss due to ineffective washout is sensitive to load challenge, as well as the nature and proportion of contaminant in the load. Thus, the lower limit of the WP region is defined by requirements of contaminant removal, while the upper limit for a given load challenge is defined by constraints of product recovery or capacity. [0072] In one or more embodiments of the invention, optimal weak partitioning conditions may be identified using the following sequence of experiments, as shown in Figure 6: (i) Perform a HTS screen (or standard batch binding experiments) to determine product partition coefficients Kp as a function of operating conditions. Identify operating window corresponding to the weak partitioning region (0.120). [0116] The supernatant from the load stage of all wells from each zone were sampled and submitted for Protein A analysis. The assay results of these samples are summarized in Table 1.1.4. There is a region of pH and conductivity where the TMAE chromatography step provides very significant removal of Protein A with limited protein loss to the resin. This region was found to be closely correlated to the partition coefficient value, Kp, and not any specific pH or chloride concentration (see Figure 8). Experiment 1.2 — Column runs under flow-through conditions [0117] The following experiment was performed in the flow-through (FT) mode, where the Mab-AAB interacts only very weakly with the column. Two runs were performed with load challenges of 110 mg/ml and 200 mg/ml of resin. 25 WO 2006/099308 PC17US2006/008919 ********[0118] For all TMAE (HiCapM) anion exchange chromatography runs described in the Series 1 experiments, the following conditions were used (exceptions are noted in the individual experimental descriptions). Operational flow rate - 150 - 300 cm/hr Equilibration 1-50 mM Tris, 2.0 M NaCl, pH 7.5 (5 column volumes) Equilibration 2 - as specified, approximately equivalent to the load pH and chloride content Post load wash - as specified, approximately equivalent to the load pH and chloride content Strip buffer - 50 mM Tris, 2.0 M NaCl, pH 7.5 (5 column volumes) Mabselect Protein A Chromato^raphy [0119] The culture containing the monoclonal antibody was purified at Pilot scale using a MabSelect column (2,389 mL) connected to a Millipore K-prime 400 chromatography system. A Mabselect Protein A column was equilibrated with 5 column volumes of 50 mM Tris/150 mM NaCl, pH 7.5 at a flow rate of 300 cm/hr. The column was then loaded at a load of approximately 40 mg product/ml resin. This was followed by a 5CV wash in 1M NaCl, 50mM Tris, pH 7.5 and a 5CV wash containing 10 mM Tris, 75mM NaCl, pH 7.5 wash. The column was then eluted using 50 mM glycine, 75mM NaCl, pH 3.0. The product pool was neutralized to pH 7.6 using 2M Tris pH 8.5. The neutralized peak had a chloride concentration of approximately 90mM. TMAE HiCap (M) Chromatographv [0120] The neutralized Protein A pool was further purified over the TMAE step with the equilibration, load, and wash solutions at pH 7.5 with 50 mM Tris and 75 mM sodium chloride. 5 column volumes of wash were used. The column dimensions and load challenges for these two studies were: Run 1: 7.0 cm diameter x 20.6 cm bed height (volume - 793 mL) with a load concentration of 11.9 mg/mL, and Run 2: 7.0 cm diameter x 13 cm bed height (volume - 500 mL) with a load concentration of 17.6 mg/mL. [0121] These load conditions were in the flow-through (FT) region (Table 1.2.1). Batch binding studies were used to measure the partition coefficient (Kp) and the bound product was determined by protein in the column strip by using UV absorbance. This method of determining the bound product typically underestimates the amount of product bound during the load due to isocratic elution of the product during the wash. The levels of Protein A, HCP and HMW in the load and product pool were measured, and the extent of removal 26 WO 2006/099308 PCT/US2006/008919 calculated. The results are presented in Table 1.2.1. There is poor removal of Protein A and HMW, and modest reduction in HCP levels. Table 1.2.1: Removal of HCP, Protein A, and HMW under flow-through conditions Run Load Challenge (mg/mL) Partition Coefficient (Kp) Bound Product. (mg/mL resin) HCP (LRV) Protein A (LRV) HMW (Fold) Recovery (%) , 1 110 0.17 1.4 2.3 0.1 - 96 2 200 0.17 3.3 2.0 <0.1 1.5 96 * Impurity levels were 38.5 ppm ProA and 51,943 ppm HCP (Run 1), 8.8 ppm ProA and 25,398 ppm HCP (Run 2). Experiment 1.3- Column runs under weak partitioning conditions (high product challenge) TMAE (HiCap M) Anion exchange Chromatography [0122] Several Mabselect Protein A runs were performed essentially as described in Experiment 1.2 to generate the load material for these runs. The partially purified antibody pool from the Protein A step was further purified over the TMAE column. The load to the TMAE column was in 50mM Tris, pH 8.2. The column diameter was 0.5 cm and the bed height was 10 cm bed height (volume - 2.0 mL). The column was challenged to a load of 500 mg/mL resin, with a load concentration of 27.7 mg/mL. [0123] The column was equilibrated with 5 column volumes of a solution containing 50 mM Tris, 2M NaC1 pH 7.5 followed by another equilibration step comprising a 50 mM Tris, pH 8.2 solution. The column was then loaded to 500 mg product/ml resin with the neutralized Protein A peak from the previous step and the product was recovered in the column effluent during the load cycle and some column volumes of the wash fraction. [0124] These load conditions are in the weak partitioning region. Batch binding studies were used to measure the partition coefficient (Kp), and product binding at high protein concentrations. At pH 8.2, and an approximate chloride content of 12 mM, the partition coefficient, Kp, is estimated to be 1.9 (from interpolation of the dataset from the HTS screen). [0125] The levels of HCP and Protein A were measured in three fractions during the loading stage representing load challenges of approximately 250, 375, and 500 mg/ml of resin. The results from experiment 1.3 are presented in Table 1.3.1. These results demonstrate that very high product challenges can be achieved in weak partitioning mode, without breakthrough of impurities. Excellent reduction in both HCP and Protein A was 27 WO 2006/099308 PCTAJS2006/008919 achieved, along with 50% reduction in HMW content. In comparison to the results for operation in the flow-through mode in Table 1.2.1, the removal of impurities was much better in the weak partitioning mode. Table 1.3.1: Removal of HCP, Protein A and HMW for a 500mg/mL TMAE load challenge Early fraction (250 mg/ml) Middle fraction (375 mg/ml) Late fraction (500 mg/ml) Final product pool (ppm) Residual HCP ppm (ng/mg product) <7.6 <7.6 <7.6 <7.6 HCP Log Removal Value (LRV) >3.5 >3.5 >3.5 >3.5 Residual Protein A ppm (ng/mg product) 0.3 Not determined 0.1 0.6 ProA Log Removal Value (LRV) 2.9 Not determined 2.3 2.5 HMW Not determined Not determined Not determined 2 fold removal * The impurities in the load were 25,398 ppm of HCP, 99.5 ppm of Protein A, and 2.3% HMW. Experiment 1.4 — Column runs under weak partitioning conditions (robustness studies) [0126] To further confirm the performance of the TMAE column in the region of weak partitioning, several runs were designed varying the pH and NaCl concentration in the load to test process robustness. All runs were performed at a load challenge of 250 mg/ml resin. Several Mabselect Protein A runs were performed essentially as described in Experiment 1.2 to generate the load material for these runs. The only factor varied in those runs was the sodium chloride concentration in the Protein A elution, which was varied to match the NaCl concentration in the TMAE load for a particular experiment. The columns were equilibrated with Equil 2 buffers and washed with Wash buffers which had approximately the same pH and sodium chloride content of the load. [0127] These load conditions are in the weak partitioning region. Batch binding studies were used to measure the partition coefficient (Kp). The runs are ranked by the partition coefficients listed in Table 1.4.1. The bound product was determined by measuring the protein in the column strip using UV absorbance, and ranges from 7.8 - 25.3 mg/mL. Protein A, HCP and HMW results from these experiments are also presented in Table 1.4.1. The removal of all impurities was found to be robust in operating ranges which cover 13.5- 38.8 mM total chloride and pH 7.8 - 8.4. 28 WO 2006/099308 PCT/US2006/008919 Table 1.4.1: Process robustness studies on removal of HCP, Protein A, and HMW in weak partitioning mode NaCl Concentration" (mM) Kp Bound Product (mg/mL) pH HCP in Load (ppm) Protein A in load (ppm) HCP (LRV) Protein A (L-RV) HMW (Fold) Recovery (%) 38.8 0.26 9.4 7.8 26,391 493.5 3.7 1.8 2.0 92 13.5 0.41 7.9 7.8 12,821 69.2 3.3 >1.9 1.8 87 27.4 0.50 8 8.0 23,465 252 3.6 2.2 3.2 91 18.5 0.73 7.8 8.0 21,626 308 3.7 >3.2 2.9 90 23.5 0.80 9.5 8.1 18,004 343 3.2 >3.2 3.5 94 27.7 0.86 95 8.2 24,821 280 3.6 >3.2 2.6 99 18.5 1.48 10 8.2 17,669 252 3.7 >3.1 3.9 95 22.0 5.35 25.3 8.4 29,293 533 3.6 >2.9 2.3 90 * Impurity levels were 38.5 ppm ProA and 51,943 ppm HCP (Run 1), 8.8 ppm PxoA and 25,398 ppm HCP (Run 2). + includes the Cl- ion contribution from NaCl, buffers and titrants Summary [0128] From this study, it can be seen that Protein A removal (LRV) varies strongly with Kp, while HCP LRV is excellent at all the values of Kp at or above 0.26, but much reduced at Kp = 0.17 (under flow-through conditions). Host cell protein removal is over one log lower for flow-through conditions compared to weak partitioning conditions, even for a reduced load challenge. The bound product ranges from 7.8 - 25 mg/mL for these weak partitioning conditions on this combination of resin and monoclonal antibody. The partition coefficient appears to be optimal between 0.4120). [0135] The supernatant from the load stage of several wells from each zone were sampled and submitted for Protein A analysis. The load had 300 ppm of Protein A. The assay results of these samples are summarized in Table 2.1.5. There is a region of pH and 31 WO 2006/099308 PCT/US2006/008919 conductivity where the TMAE chromatography step provides very significant removal of Protein A with limited protein loss to the resin. This region is found to be closely correlated to the partition coefficient value, Kp, and not any specific pH or chloride concentration. Table 2.1.5 Protein A residual levels and MAB-IMA binding data from HTS screen pH C1- (mM) Kp (predicted) Protein A (ppm) 8.5 12 42.4 <28 -BLOQ 8.5 32 4.3 <7-BLOQ 8.25 35 2.0 <5-BLOQ 8.25 45 1.13 <4-BLOQ 8.0 39 0.7 <4-BLOQ 8.25 60 0.5 <4-BLOQ 7.75 42 0.4 <4 - BLOQ 7.5 45 0.3 35 8.0 64 0.3 63 8.25 110 0.3 190 7.25 32 0.2 90 8.0 89 0.2 177 8.75 121 0.3 217 7.75 92 0.2 187 7.5 120 0.2 219 7.25 107 0.2 224 Predicted Kp values are derived from a response surface fit to the HTS screen, and subsequent prediction of the Kp based on this regression model. Experiment 2.2 - Column runs under FT conditions using TMAB-HiCapM and Mab-IMA [0136] The following experiment was performed in the flow-through (FT) mode, where the Mab-IMA interacts only very weakly with the column. Four runs were conducted, with product challenges of 109 - 275 mg/ml of resin. TMAE (HiCap M) anion exchange chromatography [0137] For all TMAE (HiCapM) anion exchange chromatography steps described in the Experiment 2 series, the following conditions were used (exceptions are noted in the individual experimental descriptions). Operational flow rate -150 - 300 cm/hr Equilibration 1-50 mM Tris, 2.0 M NaCl, pH 7.5 or 8.0 (5 column volumes) Equilibration 2 - 75mM NaCl, 50 mM Tris, pH 7.5 (runs 3 and 4 contained 50mM Glycine) Post load wash - 75mM NaCl, 50 mM Tris, pH 7.5 (runs 3 and 4 contained 50mM Glycine) Strip buffer - 50 mM Tris, 2.0 M NaCl, pH 7.5 or 8.0 (5 column volumes) 32 WO 2006/099308 PCT/US2006/008919 [0138] Several Mabselect Protein A runs were performed essentially as described in Experiment 1.2 to generate the load material for these runs. The partially purified antibody pools from the previously described Protein A step were further purified over the anion exchange step in a flow-through (FT) mode. Column diameters ranged from 1.0 - 3.2 cm and the column heights were 7.2 - 8.5 cm. [0139] The columns were equilibrated with 5 column volumes of a solution containing 50 mM Tris, 2M NaCl pH 7.5 followed by another equilibration step comprising a 50 mM Tris, pH 7.5 solution. The columns were then loaded to between 109mg/mL and 275mg/mL with the partially purified Protein A peak and the product was recovered in the column effluent during the load cycle and some column volumes of the wash fraction. [0140] These load conditions were in the flow-through (FT) region. The high throughput screen described in Experiment 2.1 provides estimates for the value of the partition coefficient (Kp) under these conditions of pH and chloride concentration. The runs are ranked by the partition coefficients listed in Table 2.2.1. The bound product was determined by measuring the protein in the column strip using UV absorbance. This method of determining the amount of bound product typically underestimates the total amount of product bound due to isocratic elution of product in the wash. Protein A, HCP, HMW and LMW removal results from these experiments are also presented in Table 2.2.1. There is relatively poor and variable removal of HCP, and no removal of Protein A and product variants (HMW and LMW species). Table 2.2.1: Removal of HCP, Protein A and HMW and LMW in FT mode Kp Produ ct bound (mg/m L) pH Cl- (mM) Loud Challenge (mg/mL) ProA In load (ppm) HCP In load (ppm) HCP (LR V) ProA (LRV) HMW (fold) LMW (fold) Recovery (%) ' 0.1 0.5 6.5 83 150 ND 4166 1.8 ND 1 1.1 >95% 0.2 0.8 7.0 83 275 25 1575 0.6 <0.l 1 1 >95% 0.2 ND 7.3 83 109 24 3117 2.4 <0.1 1 1 >95% 0.3 0.3 7.5 83 167 ND 4572 1.8 ND 1 1 >95% ND = Not determined Experiment 2.3-Column runs under weak partitioning conditions for Mab-IMA [0141] The following column experiments were performed in the weak partitioning mode under conditions identified by the HTS screening (Experiment 2.1). Seven runs were performed over the TMAE column from partially purified Protein A pools. 33 SUBSTITUTE SHEET (RULE 26) WO 2006/099308 PCT/US2006/008919 TMAE HiCap (M) chromatography [0142] The partially purified antibody from a Protein A step run essentially the same as previously described was further purified over the TMAE step under weak partitioning (WP) conditions of pH and chloride content as described below. Column diameters ranged from 0.5 to 3.2 cm and the column heights were 9.4-9.5 cm. [0143] The columns were equilibrated with 5 column volumes of a solution containing 50 mM Tris, 2M NaCl pH 7.5 or 8.0 followed by another equilibration step comprising a 50mM glycine, 50 mM Tris, pH 7.5 or 8.0 solution. The columns were then loaded to between 124mg/mL and 303mg/mL with the partially purified Protein A peak and the product was recovered in the column effluent during the load cycle and some column volumes of the wash fraction. The results from this experiment are presented in Table 2.3.1. [0144] These load conditions are in the weak partitioning (WP) region. The high throughput screen described in Experiment 2.1 provides estimates for the value of the partition coefficient (Kp). The runs are ranked by the partition coefficients listed in Table 2.3.1. The bound product was determined by measuring the protein in the column strip using UV absorbance. This method of determining the amount of bound product typically underestimates the total amount of product bound due to isocratic elution of product in the wash. Protein A, HCP, HMW and LMW results from these experiments are also presented in Table 2.2.1. There is consistent and high removal of HCP, excellent removal of Protein A, and valuable reduction of product variants (HMW and LMW species). [0145] A comparison of the data presented in Tables 2.2.1 and 2.3.1 confirms that the removal of HCP, Protein A, HMW, and LMW under conditions of a flow-through mode (Kp values of <0.3) is much lower than what can be achieved under weak partitioning conditions (Kp values >0.3), even when the load challenge exceeds 300 mg/mL. Table 2.3.1: Removal of HCP, Protein A, HMW, and LMW under weak partitioning conditions. ND = not determined Kp Product bound (mg/mL) pH Cl- (mM) Load Challenge (mg/mL) ProA In load (ppm) HCP In load (ppm) HCP (LRV) ProA (LRV) HMW (fold) LMW (fold) Recovary (%) 0.6 ND 7.5 14 303 72 754 1.9 1.6 1.3 1.5 >95% 0.7 ND 8.0 55 303 72 754 2.0 1.5 1.0 1.2 >95% 0.7 4 7.9 45 307 213 LJ852 2.6 2.4 ND ND >95% 1.0 5 8.1 45 302 222 1852 2.6 3.0 ND ND >95% 1.2 30 8.0 35 124 52 3320 2.8 >2.1 1.4 1.1 89% 1.7 ND 8.0 26 303 72 754 2.3 >2.6 1.1 1.8 >95% 1.7 9 8.1 31 310 ND L ND ND ND ND ND 90% 1.8 25 7.8 14 169 23 2462 3.0 >1.8 1.9 2.1 86% 5.2 9 8.0 17 303 72 754 2.0 >2.6 1.7 1.6 >95% 8.9 59 8.2 12 . 284 ND ND ND ND 1.5 2.1 . 75% 34 SUBSTITUTE SHEET (RULE 26) WO 2006/099308 PCT/US2006/008919 Experiments 2.4: Performance of weak partitioning column runs at Pilot and Clinical manufacturing scale [0146] The TMAE process step for the purification of Mab-IMA operated in the weak partitioning zone was scaled-up to the Pilot plant and clinical manufacturing. The culture containing the monoclonal antibody was first purified using a 3L or 5L MabSelect column in Pilot and a 28 L MabSelect column during clinical manufacturing. The MabSelect column was essentially operated as described in Experiment 1.2. The neutralized Protein A peak pools from these runs were further purified on a 1.5 L TMAE column in Pilot and a 7 L TMAE column in the clinical manufacturing facility. The results of three Pilot runs and nine clinical manufacturing runs are summarized in Tables 2.4.1 and 2.4.2, respectively. The step performance was consistent across the runs, with excellent reduction of HCP, Protein A, and good removal of product related HMW and LMW species. Product recovery was >87% in all runs. An estimate of the product bound to the resin during the Pilot runs was obtained from the product in the column strip, which ranged from 6-14 mg/mL of resin. Table 2.4.1: Performance of Pilot Scale runs under Weak Partitioning conditions ND = not determined Table 2.4.2: Performance of manufacturing scale runs under weak partitioning conditions Kp Product bound (mg/mL) pH Cl- (mM) Load Challenge (mg/mL) HCP (LRV) ProA (LRV) 11MW (fold) LMW (fold) Recovery (%) Run l 1.7 14 8.1 31 253 3.4 >2.6 2.0 2.0 90 Run 2 1.7 13 8.1 31 184 >3.6 >2.6 1.0 3.0 88 Run 3 1.7 6 8.1 31 150 ND >2.8 1.3 1.2 88 Kp Load Challenge (mg/mL) Cl- (mM) pH HCP* (LRV) ProA* (LRV) HMW (fold) LMW (fold) Recovery (%) Run 1 3.2 90 30 8.1 >2.2 ≥0.8 2.5 3.0 90 Run 2 2.5 180 29 8.0 >2.2 ≥0.0 1.9 1.5 93 Run 3 2.6 133 28 8.0 >2.3 ≥1.0 2.5 1.8 95 Run 4 2.6 174 29 8.0 >2.2 ≥1.0 2.5 1.2 94 35 SUBSTITUTE SHEET (RULE 26) WO 2006/099308 PCT/US2006/008919 Run5 2.2 136 31 8.0 >2.0 ≥0.9 2.3 1.7 102 Run 6 3.8 146 27 8.1 >2.2 ≥0.8 2.5 1.8 91 Run 7 2.0 118 28 7.9 >2.2 ≥1.0 1.6 1.6 96 * The HCP and Protein A levels in TMAE peak pool were below limit of quantitation. Summary [0147] HTS identified conditions for WP and FT operation. The FT mode provided only a modest reduction in HCP and LMW species, and no reduction in Protein A residuals or HMW species. Operation in the WP mode improves the removal of all impurities without sacrificing product yield. The process step was scaled up to the Pilot plant and operated consistently for three runs, with very high LRVs for HCP and Protein A, and good reductions in HMW and LMW species. Series 3 - Anion Exchange using TMAE-HiCapM and Mab-AAB Experiment 3.1 — High throughput screen to establish WP and FT conditions [0148] Experiment 3.1 was performed using procedures as described in Experiment 1.1. Experiment 3.2 — Column capacity runs under conditions corresponding to varying partition coefficients [0149] Five chromatography experiments were performed under conditions corresponding to a range of partition coefficients identified by HTS screen (Experiment 3.1). The TMAE columns were loaded to a very high load challenge (>1000 g/L) to specifically highlight the superior performance of the AEX step under weak partitioning conditions. [0150] The following conditions were used for the AEX runs performed in Series 3 (exceptions are noted in the individual experimental descriptions). Operational flow rate -150 - 300 cm/hr Equilibration 1 - 50 mM Tris, 2.0 M NaCl, pH 7.5 (5 column volumes) Equilibration 2 - as specified, approximately equivalent to the load pH and chloride content Post load wash- as specified, approximately equivalent to the load pH and chloride content Strip buffer - 50 mM Tris, 2.0 M NaCl, pH 7.5 (5 column volumes) [0151] The column was equilibrated with 5 column volumes of equilibration buffer 1 followed by 5 column volumes of equilibration 2 step. The column was then loaded to 36 WO 2006/099308 PCT/US2006/008919 between 940 and 2144 mg of product /ml of resin with the Protein A peak pool (refer to Series 1, Experiment 1.1) adjusted to the appropriate equilibration 2 buffer. [0152] The column effluent fractions were collected and subsequently assayed for HCP and residual Protein A levels. The load conditions used in these experiments correspond to progressively increasing partition coefficients that span the flow-through and weak partitioning (WP) regions. The high throughput screen described in Experiment 3.1 provided estimates for the value of the partition coefficient (Kp). The bound product values in this example were calculated based on the product eluted in the strip. The results from these experiments are presented in Table 3.2.1 and Figures 10A and 10B. Table 3.2.1 - Summary of results from very high load challenge experiments in the WP mode Partition Coefficient Kp Operating conditions Load Challenge mg/ml Product bound mg/mL* Recovery* Run 1 0.1 Flow-through 1754 0 100 Run 2 0.23 Weak Partitioning 940 14.2 98.5 Run 3 0.8 Weak Partitioning 940 12.0 98.7 Run 4 0.8 Weak Partitioning 2144 23.0 L98.9 Run 5 2.73 Weak Partitioning 960 12.6 98.7 Run 6 7 Weak Partitioning 1130 71.7 93.7 *Based on mass balance calculations. [0153] The product bound value for the ran corresponding to a Kp of 0.1 was near zero, as is expected for a typical flow-through operation. The product bound values for experiments performed in the weak partitioning region were > 12.0 mg/ml in all cases. In fact, the product bound value for the run corresponding to the Kp of 7 was as high as 71 mg/ml. The product recovery in the combined load eluate and wash fractions in all cases were, however, > 93%. [0154] HCP and Protein A removal, as a function of load challenge, is presented in Figures 10A and 10B. As discussed earlier, the HCP removal increases significantly as conditions move from flow-through to weak partitioning. Operating under flow-through conditions provides approximately 1.5 logs of HCP clearance, while the HCP log removal values were as high as 3.8 logs at load challenges < 450 mg /ml of resin when operated at a Kp of 7 in the weak partitioning region. At a Kp of 0.8 in the weak partitioning region, 2.8 logs of HCP clearance was obtained for load challenges up to 1000 mg/ml of the resin, and > 3 logs of HCP clearance was obtained for up to a load challenge of 800 mg/ml of resin at a Kp of 2.7 in the weak partitioning region. 37 WO 2006/099308 PCT/US2006/008919 [0155] As in the case of HCP, Protein A removal increases significantly as we move from flow-through conditions to weak partitioning conditions. The results presented in Figures 10A and 10B also highlight the fact that increasing the Kp of operation from the flow-through to weak partitioning region increases both the HCP and Protein A log clearance values obtained prior to breakthrough of the contaminant, as well as the load challenge corresponding to the point of breakthrough. A further increase in Kp continues to increase the HCP and Protein A LRV prior to breakthrough of the contaminant. However, the point of breakthrough occurs at relatively lower load challenges as the bound product now competes with the contaminants for the binding sites. Nevertheless, the column capacity for the runs presented here were very high even for the high Kp run. Summary [0156] In this example it was shown that Protein A and HCP removal can be significantly improved by operating the AEX step under weak partitioning conditions and at load challenges in excess of 1000 mg/ml of resin. This example highlights one fundamental difference between weak partitioning chromatography and the standard operations under binding conditions. The weak partitioning conditions push the limits of product binding only up to a point where the contaminant clearance is significantly improved, while product recovery and load challenges remain high. The Kp values corresponding to binding conditions are > 20 in AEX; under these conditions the competitive effects between product and contaminant are very strong leading to reduced capacity as compared to weak partitioning chromatography. Series 4- Hvdrophobic Interaction using Phenyl Toyopearl and Mab-AAB Experiment 4.1 - Batch binding studies to establish WP and FT conditions [0157] Batch binding studies were conducted to identify the weak partitioning and flow-through conditions for Mab-AAB with Phenyl Toyopearl medium from Tosoh Biosciences. The salt modulating the strength of the product interaction with the resin is Na2SO4, which was varied from 0.20 to 0.90M. The solutions were buffered to control at pH 7.5. 45 um filter plates were used to incubate the resin with liquid and to decant the supernatant through centrifugation. Eight Tris/Na2SO4 buffers were made with Na2SO4 at different concentrations (0.2 M to 0.9 M). Mab-AAB which was partially purified by Protein A chromatography was diluted into Tris/ Na2SO4 solution to a final of concentration of 0.87 mg/ml. 50 ul of resin was equilibrated with 300 ul of buffer and then the supernatant 38 WO 2006/099308 PCT/US2006/008919 decanted for each of the Tris/ Na2SO4 conditions; this equilibration was repeated three times. After equilibration, decanted resiri was mixed with product at the same salt concentration and pH and incubated for 30 minutes with gentle shaking. The load challenge was 5.2 mg product /ml of resin for all conditions. A UV plate was then stacked at the bottom of the filter plate to collect the supernatant upon centrifiigation. Subsequently, 300 ul of 50mM Tris, pH 7.5 buffer was applied to the resin to strip off bound product. Following a 20 minute incubation, the strip was collected in a separate UV plate through centrifugation. The concentration of the product in each fraction was measured by UV absorbance and the extinction coefficient for this MAb. The calculations were adjusted for a stage-to-stage carry over volume of 29 ul that was determined through a separate set of experiments. The experiment was repeated four times under each salt condition and an average partition coefficient is reported. [0158] Table 4.2.1 summarizes the partition coefficients from this experiment. The highest concentrations of Na2SO4 caused strong product binding, while salt concentrations in the range of 0.40 - 0.55M represent weak partitioning conditions. Experiment 4.2 - Column runs under weak partitioning, flow-through and binding conditions (high product challenge studies) [0159] Column runs were performed under flow-through, weak partitioning and strong binding conditions. For all Phenyl Toyopearl hydrophobic interaction chromatography runs described in the Series 4 experiments, the following conditions were used (exceptions are noted in the individual experimental descriptions). Column dimension: diameter 0.5 cm, bed height 9.5 - 10.5 cm Equilibration - 50 mM Tris, pH 7.5 with [Na2SO4] approximately equivalent to the load Load - [Na2SO4] as specified below Wash - [Na2SO4] equal to the load (exceptions noted below) Strip- 50 mM Tris, pH 7.5 [0160] Two different loads were used: i) partially purified antibody pools from a Protein A step run essentially the same as those previously described or ii) more pure TMAE Q Sepharose FF product pools from FT mode operation. Experiment 4.2.1 - Column runs using Protein A peak pool as load [0161] The experiments discussed here were performed to highlight the superior performance of HIC under weak partitioning conditions. Column runs were performed under 39 WO 2006/099308 PCT/US2006/008919 varying salt concentrations to cover a range of partition coefficients that correspond to flow- through, weak partitioning and strong binding conditions. The batch binding screen described in Experiment 4.1 provides estimates for the value of the partition coefficient (Kp). The columns were equilibrated with 5 column volumes of a solution containing 50 mM Tris, pH 7.5 and appropriate [Na2SO4] at specified concentration. The Protein A peak was first concentrated 10-fold, and subsequently diluted to 14.77 mg/ml in the appropriate salt concentration. Host cell protein (HCP) and residual Protein A levels in the load material were 30911 ppm and 17.1 ppm, respectively. All column runs were performed at a load challenge of 100 mg/ml of resin. The product was collected in the column effluent during the load cycle fraction. After product flow-through, ten column volumes of wash buffer at the same salt concentration as load were applied to the column, followed by five column volumes of a strip buffer containing 50mM Tris.pH 7.5. HCP and Protein A content in the load eluate and wash samples were subsequently analyzed by ELISA. The combined impurity level in both load eluate and wash fractions is reported in Table 4.2.1. [0162] The runs are ranked by the partition coefficients. The bound product was determined by measuring the protein in the column strip using UV absorbance. This method of determining the bound product typically underestimates the amount of product bound during the load due to the gradual desorption of the product during the wash. Table 4.2.1: Impurity removal in flow-through, weak partitioning and strong binding conditions Na2SO4 Cone Partition Coefficient Kp Operating window Bound Product (mg/mL) Product Recovery (%) Protein A removal (fold) HCP removal (Log) Run 1 0.10 M <0.1 Row-Through 0.6 94 0.9 0.3 Run 2 0.20 M <0.1 Flow-Through 1.3 93 0.8 0.4 Run 3 0.40 M 0.9 Weak Partitioning 2.8 94 1.7 1.0 Run 4 0.45 M 2.0 Weak Partitioning 3.0 93 1.5 0.9 Run 5 0.50 M 4.3 Weak Partitioning 3.8 92 2.5 N/A Run 6 0.55 M 9.9 Weak Partitioning 5.0 93 3.4 1.1 Run 7 0.80 M >100 Strong Binding 25 72 2.2 0.7 Run 8 0.90 M >100 Strong Binding 34 67 1.1 0.4 (The partition coefficient Kp accounts for the phase volume ratio of 6 from 50 microliters of resin and 300 microliters of solution.) [0163] As is evident from the data presented in Table 4.2.1, the performance of the HIC step improves significantly with respect to contaminant reduction as we move from flow-through conditions to weak partitioning conditions, while product recovery is comparable. A further increase in the operating salt concentration leads to partition coefficients that correspond to the strong binding conditions. It is once again clear from the 40 SUBSTITUTE SHEET (RULE 26) WO 2006/099308 PCT/US2006/008919 data presented in Table 4.2.1 that the HIC step performance deteriorates with respect to both contaminant reduction, as well as product recovery, as we move from weak partitioning to strong binding conditions. The optimum operating window for this separation therefore corresponds to that of weak partitioning chromatography. Under weak partitioning conditions, the HIC step provides 1 log reduction of HCP and a 3.4 fold reduction of Protein A. The bound product levels under the weak partitioning conditions, in this example, were between 2.8 ~ 5 mg/ml of the resin. Experiment 4.2.2 - Column runs using Q-Sepharose peak pool as load [0164] The Q Sepharose FF peak pool was used in these sets of experiments to highlight the fact that the performance of the HIC step under the optimum weak partitioning chromatography conditions can be further improved with a cleaner feedstock. The load material in this case contained 2880 ppm of HCP and was generated by purifying the Protein A peak pool on a Q-Sepharose FF column. Two experiments, one under weak partitioning conditions and the other under typical flow-through conditions, were conducted to compare the column performance with respect to impurity removal. The Q-Sepharose peak was diluted to 3.27 mg/ml at 550 mM Na2SO4 and loaded to the column to a load challenge of 100 mg/ml of resin for operation under weak partitioning conditions. The column was subsequently washed with 10 CVs of a buffer containing the same salt concentration as the load and stripped with 6CV of 50 mM Tris buffer, pH 7.5. The second experiment was conducted under flow-through conditions. The load was adjusted to 3.03 mg/ml in 200 Na2SO4 and loaded to the column to a load challenge of 90 mg/ml of resin. The column was then washed with 6CV of a buffer containing the same salt concentration as the load, and subsequently stripped with 6 CV of 50 mM Tris buffer, pH 7.5. In both runs, the flow- through and wash fractions were collected for recovery and impurity analysis. The results from these runs are reported in Table 4.2.2. Table 4.2.2: Comparison of HIC weak partitioning results to flow-through results Run Na2SO4 Cone Partition Coefficient (Kp) Operating mode Load Challenge (mg/mL) Bound Product (mg/mL) Product Recovery HCP LRV 1 0.55M 9.9 Weak Partitioning 100 3.0 94% >2 2 0.20M <0.1 Flow-through 90 0.1 99% <0.5 [0165] The product recovery values under weak partitioning conditions were comparable to flow-through operations and were also independent of the feedstock used in these experiments. The performance of the steps with respect to HCP removal is 41 SUBSTITUTE SHEET (RULE 26) WO 2006/099308 PCT/US2006/008919 significantly higher under weak partitioning conditions as compared to flow-through operation. [0166] HCP LRV across the HIC step with either feedstock was comparable under flow-through conditions (~ 0.4 - 0.5 LRV). However, the HCP LRV values for experiments performed under weak partitioning conditions increased from 1 LRV with the Protein A load material to greater than 2 LRV with load material purified through Protein A and Q Sepharose FF columns. Summary [0167] Another mode of chromatography (HIC) was shown to operate successfully in a weak partitioning mode. The performance of the HIC step under weak partitioning conditions was shown to be superior to both flow-through conditions, as well as to operations under tighter binding conditions, with respect to product recovery and HCP/Protein A removal. A high load challenge capacity of 100 mg/ml of resin was successfully processed under weak partitioning conditions, where > 3 mg/mL of product bound to the resin (even though the load concentration was 3.27 mg/mL). The partition coefficients corresponding to optimum weak partitioning conditions appear to be slightly higher than those for anion exchange chromatography. Series 5- Hydroxyapatite using ceramic Hydroxyapatite Type I and Mab-MYA Experiment 5.1 - High throughput screen to establish WP and FT conditions [0168] A high throughput screen (HTS) was performed to identify the weak partitioning and flow-through conditions for Mab-MYA with ceramic hydroxyapatite medium. This screen varied the concentration of sodium chloride and sodium phosphate to determine their effect on the extent of binding of MAB-MYA to the hydroxyapatite medium. [0169] 50µL of ceramic hydroxyapatite medium was added to 30 wells of a 96 well filter plate. Each well was equilibrated in solutions made up of the appropriate sodium chloride and sodium phosphate concentrations in a 100mM HEPES buffer containing 100mM arginine at pH 7.2. The concentrations of the two salts in the solution are shown in Tables 5.1.1 and 5.1.2. Each condition was performed in duplicate. The MAB-MYA load challenge in each of these wells was of 5.0 mg/mL of resin 42 WO 2006/099308 PCT/US2006/008919 Tabie 5.1.1: Sodium chloride levels in each well (in mM) 1 2 3 4 A 50 200 50 200 B 750 50 750 50 C 50 380 50 380 D 500 760 500 760 E 50 1140 50 1140 F 200 200 200 200 G 350 200 350 200 H 700 700 Table 5.1.2: Sodium phosphate levels in each well (in mM) 1 2 3 4 A 5 20 5 20 B 5 30 5 30 C 8 30 8 30 D 8 30 8 30 E 10 30 10 30 F 10 50 10 50 G 10 100 10 100 H 10 10 [0170] In the first stage of the HTS experiment, each well was equilibrated in the conditions of sodium chloride and sodium phosphate as described in Tables 5.1.1 and 5.1.2, in a phase volume ratio of 6:1 (300µL solution: 50uL resin). The plate was shaken for 20 minutes, allowing equilibrium to be reached. The solution was then removed by centrifuging the filter plate. This equilibration cycle was repeated three times. [0171] In the second stage, the resin in each well was challenged with a concentrated MAb-MYA solution to the appropriate protein load challenge with a volume ratio of 6:1 (300µL solution: 50uL resin) at the appropriate sodium chloride and sodium phosphate concentration. A 7.0 mg/mL solution of Mab-MYA in 50 mM NaCl, 100 mM HEPES, 100 mM arginine, pH 7.2 was used as stock solution. The loaded plate was shaken for 20 minutes, allowing the resin and solution to equilibriate. The supernatant was removed from the filter plate by centrifugation and collected into a collection plate. The protein concentration in the supernatant in each well was determined by absorbance at A280nm. [0172] In the third stage, resin was washed by adding solutions of the specified sodium chloride and sodium phosphate conditions listed in Tables 5.1.1 and 5.1.2. The supernatant was removed after shaking for 20 minutes. 43 WO 2006/099308 PCT/US2006/008919 1 [0173] In the fourth stage, a buffer comprised of 100mM sodium phosphate, 1M NaCl pH 7.2 was added to remove the remaining protein that was bound to the resin. [0174] The partition coefficients were calculated for each well using the mass eluted from stage 4 and the product concentration from stage 2, and are shown in Table 5.1.3. Table 5.1.3: Partition Coefficients (Kp) for the 96 well HTS screen for MAB-MYO 1 2 3 4 A 49.3 2.3 50.5 2.4 B 3.5 6.0 4.1 6.0 C 31.6 0.4 34.7 0.3 D 2.9 0.1 3.3 0.1 E 28.1 0.0 28.3 0.0 F 7.2 0.5 7.7 0.4 G 3.5 0.0 3.1 0.0 H 1.1 1.1 [0175] As shown in Table 5.1.3, the Kp value can be used to describe regions where MAB-MYA binds to the hydroxyapatite medium with different strengths. The strength of MAB-MYA binding to ceramic hydroxyapatite medium can be manipulated by varying conditions of chloride and phosphate concentration into flow-through (Kp=<0.1), weak partitioning (0.120). Experiment 5.2- Column runs under WP conditions [0176] The experiments discussed here were specifically performed to highlight the superior performance of the cHA step under weak partitioning conditions. The experiments were therefore performed under conditions corresponding to a range of partitioning coefficients identified by the HTS screen (Experiment 5.1). Twelve runs were conducted, with product load challenges of 100 mg/ml of resin. Mabselect Protein A Chromatography [0177] The culture containing the monoclonal antibody was purified using a MabSelect column. A Mabselect Protein A column was equilibrated with 5 column volumes of 50 mM Tris/150 mM NaCl, pH 7.5 at a flow rate of 300 cm/hr. The column was then loaded at a load of approximately 40 mg product/ml resin. This was followed by a 10CV 44 WO 2006/099308 PCT/US2006/008919 wash in 1M arginine, 50mM Tris, pH 7.5 and a 5CV wash containing 10 mM Tris, 75mM NaCl, pH 7.5 wash. The column was then eluted using 100mM arginine, 50mM NaCl, pH 3.0. The product pool was neutralized to pH 7.2 using 2M HEPES pH 8.0. Ceramic Hydroxyapatite Chromatography [0178] The partially purified antibody pools from the Protein A step were further purified over hydroxyapatite. The column diameter was 0.5 cm and the column height was 10 cm. [0179] For all hydroxyapatite chromatography steps described in the Experiment 5 series, the following conditions were used (exceptions are noted in the individual experimental descriptions). Operational flow rate - 150 - 240 cm/hr Equilibration 1 300 mM sdium posphate, 1.0M NaCl, pH 6.8 (3 column volumes) Equilibration 2 5-30 mM sodium phosphate, 50 - 760 mM NaCl, 100mM Arg, 100mM HEPES pH 7.2 (5 column volumes) Wash 5 - 30 mM sodium phosphate, 50 - 760 mM NaCl, 100mM Arg, 100mM HEPES pH 7.2 (5 -10 column volumes) [0180] The column was equilibrated with 5 column volumes of equilibration buffer 1 followed by another equilibration 2 step. The column was then loaded to 100 mg product/ml resin with the Protein A peak from the previous step (adjusted to the appropriate equilibration 2 buffer), and the product was recovered in the column effluent during the load cycle and some column volumes of the wash fraction. The results from these experiments are presented in Table 5.2.1 and Figure 11. [0181] These load conditions were in the flow-through, weak partitioning (WP) and binding regions. The high throughput screen described in Experiment 5.1 provides estimates for the value of the partition coefficient (Kp) and the bound product (mg/mL of resin) under these conditions of chloride and phosphate concentration. The bound product was determined from the product breakthrough volumes from the column runs. HCP and Protein A results from these experiments are presented in Table 5.2.1 and Figure 11. Table 5.2.1: Removal of HCP and Protein A under flow-through, weak partitioning and binding conditions Partition Coefficie nt Kp Operating Mode Phos NaCl Bound Product (mg/mL) Product Recovery (%) Protein A removal (fold) Host Cell Protein (LRV) 45 WO 2006/099308 PCT/US2006/008919 Run 1 0.0 Flow-through 30 760 0 93 0 0.58 Run 2 0.9 Weak Partitioning 30 200 3.0 96 NA 0.57 Run 3 1.1 Weak Partitioning 10 700 3.2 94 1.7 0.9 Run 4 1.7 Weak Partitioning 20 200 3.1 94 NA 0.83 Run 5 3.0 Weak Partitioning 8 500 3.0 100 1.9 1.3 Run 6 3.2 Weak Partitioning 10 350 3.6 94 2.1 1.5 Run? 3.7 Weak Partitioning 5 750 3.3 99 2.4 1.2 Run 8 5.8 Weak Partitioning 30 50 10.7 95 2.1 1.2 Run 9 7.3 Weak Partitioning 10 200 10.2 94 2.4 1.2 Run 10 27.8 Strong Binding 10 50 16.3 91 2.1 1.2 Run 11 32.7 Strong Binding 8 50 21.6 86 2.4 1.4 Run 12 49.2 Strong Binding 5 50 24.6 79 2.1 1.6 [0182] It is evident from the data presented in Table 5.2.1 and Figure 11 that the performance of the cHA step improves significantly with respect to contaminant reduction as we move from flow-through conditions to weak partitioning conditions, while product recovery is comparable. Operating under conditions corresponding to a further increase in the partition coefficient (i.e., operating in the binding region) provides no additional benefit with respect to contaminant removal. However, product recovery across the step begins to drop under strong binding conditions. Thus, the optimum operating window for this separation corresponds to that of weak partitioning chromatography. Under these conditions, > 2 log reduction of Protein A and > 1.2 log reduction of host cell protein was obtained at a load challenge of 100 mg of product / ml of resin. Bound product levels under the weak partitioning conditions, in this example, were between 3.0-10.2 mg/mL of the resin. Summary [0183] A third mode of chromatography (hydroxyapatite) was shown to operate successfully in a weak partitioning mode. Protein A and HCP bind more tightly than the product antibody to ceramic resin, and are retained strongly under WP conditions. Higher values of Kp in the WP region are between 10 and 20 in some cases, which still provide good product recovery (> 90%). Lower levels of Kp give correspondingly higher recoveries. [0184] In this example it was shown that the performance of the column step can be optimized primarily through the choice of partition coefficient used to run the column. The partition coefficient in hydroxyapatite is a complex function of pH, salt (type and concentration), phosphate, and buffer components. All of these variables in general have an impact on the performance of the column step. The approach presented here provides a simple means of relating the impact of changing any one of these variables on column performance. The unified 'partition coefficient' approach presented in this example opens up the possibility of operating in a wider operating space in this mode of chromatography than 46 WO 2006/099308 PCT/US2006/008919 has been done before. The weak partitioning conditions for optimum performance can easily be identified using the HTS methods described above. Series 6- Hydroxyapatite using ceramic Hydroxyapatite Type I and Mab-A5T Experiment 6.1 - High throughput screen to establish WP and FT conditions [0185] A high throughput screen (HTS) was performed to identify the weak partitioning and flow-through conditions for Mab-A5T with ceramic hydroxyapatite medium. This screen varied pH, sodium chloride and sodium phosphate concentrations to determine their effect on the extent of binding of MAB-A5T to the hydroxyapatite medium. [0186] 50µL of ceramic hydroxyapatite medium was added to 36 wells of a 96 well filter plate. Each well was equilibrated in solutions made up of the appropriate sodium chloride and sodium phosphate concentrations in a 50mM HEPES buffer containing 50mM arginine at either pH 7.0 or pH 8.0. The concentrations of the two salts in the solution are shown in Tables 6.1.1 and 6.1.2. The conditions shown in columns 1-3 were performed at pH 7.0, and columns 4-6 were performed at pH 8.0. The MAB-A5T load challenge in each of these wells was of 5.0 mg/mL of resin. Table 6.1.1: Sodium chloride levels in each well (in mM) 1 2 3 4 6 pH7.0 pH8.0 A 50 400 400 50 400 400 B 50 50 400 50 50 400 C 50 50 50 50 D 100 50 100 50 E 100 100 100 100 F 100 100 100 100 G 400 100 400 100 H 400 400 400 400 Table 6.1.2: Sodium phosphate levels in each well (in mM) 1 2 3 4 5 6 pH7.0 pH8.0 A 2 32 8 2 32 8 B 8 2 32 8 2 32 C 32 8 32 8 D 2 32 2 32 E 8 2 8 2 F 32 8 32 8 G 2 32 2 32 H 8 2 8 2 47 WO 2006/099308 PCT/US2006/008919 [0187] In the first stage of the HTS experiment, each well was equilibrated in the conditions of sodium chloride, sodium phosphate and pH as described in Tables 6.1.1 and 6.1.2 in a phase volume ratio of 6:1 (300µL solution: 50uL resin). The plate was shaken for 20 minutes, allowing equilibrium to be reached. The solution was then removed by centrifuging the filter plate. This equilibration cycle was repeated three times. [0188] In the second stage, the resin in each well was challenged with a concentrated MAb-A5T solution to the appropriate protein load challenge with a volume ratio of 6:1 (300µL solution: 50uL resin) at the appropriate pH and sodium chloride and sodium phosphate concentration. A 6.9 mg/mL solution of Mab-A5T in 1mM HEPES, 100 mM NaCl, pH 7.0 was used as stock solution. The loaded plate was shaken for 20 minutes, allowing the resin and solution to equilibrate. The supernatant was removed from the filter plate by centrifugation and collected into a collection plate. The protein concentration in the supernatant in each well was determined by absorbance at A280nm. [0189] In the third stage, resin was washed by adding solutions of the specified sodium chloride, sodium phosphate and pH conditions listed in Tables 6.1.1 and 6.1.2. The supernatant was removed after shaking for 20 minutes. [0190] In the fourth stage, a buffer comprising 100mM sodium phosphate, 1M NaCl pH 7.2 was added to remove the remaining protein that was bound to the resin. The partition coefficients were calculated for each well using the mass eluted from stage 4 and the product concentration from stage 2, and are shown in Table 6.1.3. Table 6.1.3: Partition Coefficients (Kp) for the HTS screen for MAB-A5T 1 2 3 4 5 6 pH7.0 pH8.0 A 142.7 0.1 1.6 50.2 0.0 0.3 B 90.6 144.3 0.1 9.9 44.7 0.0 C 12.1 84.5 1.0 13.7 D 90.5 10.4 22.2 1.1 E 27.5 94.1 4.3 21.9 F 2.5 28.0 0.3 4.3 G 15.1 2.1 2.2 0.4 H 1.2 15.1 0.4 2.0 [0191] As shown in Table 6.1.3, the Kp value can be used to identify regions where MAB-A5T binds to the hydroxyapatite medium with different strengths. The strength of MAB-A5T binding to ceramic hydroxyapatite medium can be manipulated by varying 48 WO 2006/099308 PCT/US2006/008919 conditions of NaCl, phosphate and pH into flow-through, weak partitioning, and binding zones. Experiment 6.2 — Column runs under WP conditions [0192] Experiments were performed to highlight the superior performance of the cHA step under weak partitioning conditions. The experiments were therefore performed under conditions corresponding to a range of partitioning coefficients identified by the HTS screen (Experiment 6.1). Eight runs were conducted, with product load challenges of 110 mg/ml of resin. Mabselect Protein A Chromatography [0193] The culture containing the monoclonal antibody was purified using a MabSelect column. A Mabselect Protein A column was equilibrated with 5 column volumes of 50 mM Tris/150 mM NaCl, pH 7.5 at a flow rate of 300 cm/hr. The column was then loaded at a load challenge of approximately 40 mg product/ml resin. This was followed by a 5CV wash in 1M NaCl, 50mM Tris, pH 7.5 and a 5CV wash containing 10 mM Tris, 75mM Nacl, pH 7.5 wash. The column was then eluted using 100mM arginine, 50mM NaCl, pH 3.0. The product pool was neutralized to pH 7.2 using 2M HEPES pH 8.0. Ceramic Hydroxyapatite Chromatography [0194] The partially purified antibody pools from the Protein A step were further purified over hydroxyapatite. The column diameter was 0.5 cm and the column height was 10 cm. For all hydroxyapatite chromatography steps described in the Experiment 6 series, the following conditions were used (exceptions are noted in the individual experimental descriptions). Operational flow rate -150 - 240 cm/hr Equilibration 1 300 mM sodium phosphate, 1.0M NaCl, pH 6.8 (3 column volumes) Equilibration 2 2-32 mM sodium phosphate, 50 - 400 mM NaCl, 5 mM Imidazole, 50 mM glycine, 10 mM HEPES, pH 7.0 (5 column volumes) Wash Same as Equilibration 2. 49 WO 2006/099308 PCT/US2006/008919 [0195] The column was equilibrated with 5 column volumes of equilibration buffer 1 followed by another equilibration 2 step. The column was then loaded to 110 mg product/ml resin with the Protein A peak from the previous step (adjusted to the appropriate equilibration 2 buffer), and the product was recovered in the column effluent during the load cycle and some column volumes of the wash fraction. The results from these experiments are presented in Table 6.2.1 and Figure 12. Table 6.2.1: Partition coefficients for MAB-A5T on cHA resin and the corresponding operating window. Partition Coefficient Kp Product Bound mg/ml Operating mode Phos mM NaCl mM PH Run1 0.1 0 Flow-Through 32 400 7.0 Run 2 0.7 1.6 Weak Partitioning 32 170 7.0 Run 3 1.4 2.2 Weak Partitioning 32 120 7.0 Run 4 2.1 1.6 Weak Partitioning 2 400 7.0 Run 5 13.7 7.0 Weak Partitioning 8 50 8.0 Run 6 22 6.7 Weak Partitioning 2 100 8.0 Run 7 54 12.8 Strong Binding 2 60 8.0 Run 8 >100 16 Strong Binding 2 50 7.0 [0196] The operating conditions in these experiments correspond to the flow-through, weak partitioning (WP) region and binding regions. The HTS experiment described in Experiment 6.1 provides estimates for the value of the partition coefficient (Kp) under these conditions of pH, chloride and phosphate concentrations. The runs in Table 6.2.1 are ranked by the partition coefficients. The bound product was determined by measuring the protein in the column strip using UV absorbance. This method of determining the bound product typically underestimates the amount of product bound during the load due to the gradual desorption of the product during the wash. HCP and product related HMW removal, as well as product recovery results from these experiments are presented in Figure 12. [0197] It is clear from the data presented in Figure 12 that the performance of the cHA step improves significantly with respect to HCP and HMW reduction as we move from the flow-through conditions to the weak partitioning conditions, while product recovery is maintained at > 80%. Operating under conditions corresponding to a further increase in the partition coefficient (i.e., operating in the binding region) provides no additional benefit with respect to contaminant removal. However, the product recovery across the step begins to drop under strong binding conditions. Thus, the optimum operating window for this separation corresponds to that of weak partitioning chromatography. Under these conditions, 50 WO 2006/099308 PCT/US2006/008919 a 4-fold reduction of product related HMW species and >1.4 log reduction of HCP was obtained at a load challenge of 110 mg of product/ml of resin. The bound product levels under weak partitioning conditions, in this example, were between 1.6-6.7 mg/ml of the resin. Summary [0198] A second example was presented in hydroxyapatite where operating under weak partitioning chromatography was shown to provide improved performance with respect to HCP and HMW reduction and product recovery (> 80%). As in previous examples, the performance of the step was optimized primarily through the choice of partition coefficients used to run the column. The approach presented here provides a simple means of relating the impact of changing any one of several variables (pH, salt, phosphate, imidazole, glycine, HEPES, etc.,) to column performance. The weak partitioning conditions for optimum performance can easily be identified using the HTS methods described in this example. The approach presented here opens up the possibility of operating in a wider operating space in this mode of chromatography than has been done before. The optimal WP region in this example corresponds to partition coefficients between 2 and 20. Series 7- Hydroxyapatite using Ceramic Hydroxyapatite Type I and Mab-MYO Experiment 7.1 - High throughput screen to establish FT, WP and strong binding conditions [0199] A high throughput screen (HTS) was performed to identify flow-through, weak partitioning and binding conditions for Mab-MYO with ceramic hydroxyapatite medium. This screen varied the concentration of pH, arginine/glycine, HEPES, sodium phosphate and sodium chloride to determine their effect on the extent of binding of MAB- MYO to the hydroxyapatite medium. [0200] The HTS procedures used in this example were similar those described in Series 5 and Series 6 and are not discussed here. Predicted Kp values derived from a response surface fit to the HTS data were used to pick specific conditions for column experiments. Experiment 7.2 - Column runs under WP conditions [0201] The experiments discussed here were performed under conditions corresponding to a range of partitioning coefficients identified by the HTS experiments. These experiments were specifically performed to highlight the superior performance of the 51 WO 2006/099308 PCT/US2006/008919 cHA step under weak partitioning conditions for the removal of HCP and product related HMW species. Four runs were conducted with a product load challenges of 55 mg/ml of resin. The load challenge used in these experiments is low for weak partitioning chromatography, but is typical for flow-through operation. No attempt was made in these experiments to optimize load challenge for weak partitioning chromatography. [0202] The partially purified antibody pools from the Protein A step were used in these experiments. The column diameter was 0.5 cm and the column height was 10 cm. [0203] For all hydroxyapatite chromatography steps described in the Experiment 7 series, the following conditions were used (exceptions are noted in the individual experimental descriptions). Operational flow rate -150 - 240 cm/hr Equilibration 1 300 mM sodium phosphate, 1.0M NaCl, pH 6.8 (2-5 column volumes) Equilibration 2 1-8 mM sodium phosphate, 50 - 1750 mM NaCl, 12 - 50 mM Arg, 20-50mM HEPES pH 7.0 (5 column volumes) Wash Same as Equilibration 2 [0204] The column was equilibrated with 2-5 column volumes of equilibration buffer 1 followed by 5 column volumes of equilibration 2. The column was then loaded to 55 mg product/ml resin with the Protein A peak (adjusted to the appropriate equilibration 2 buffer), and the product was recovered in the column effluent during the load cycle and some column volumes of the wash fraction. The results from these experiments are presented in Table 7.2.1 and Figure 13. Table 7.2.1: Partition coefficients for MAB-MYO on cHA resin and the corresponding operating mode Partition Coefficient Kp Operating mode Arg mM Hepes mM Phos mM NaCl mM Product Bound mg/ml Run1 3.6 Weak Partitioning 50 20 8 300 5.1 Run 2 4.2 Weak Partitioning 50 20 2 600 8.2 Run 3 8.9* Weak Partitioning 12 20 1 1750 9.5 Run 4 >100 Strong Binding 50 50 5 50 41.6 * Optimal Kp condition, see Figure 13 [0205] The operating conditions in these experiments correspond to the flow-through, weak partitioning (WP) and binding regions. The HTS experiment described in Experiment 7.1 provides estimates for the value of the partition coefficient (Kp) under these conditions of pH, chloride, phosphate, glycine / arginine and HEPES concentration. The runs in Table 52 WO 2006/099308 PCT/US2006/008919 7.2.1 are ranked by the partition coefficients. The bound product was determined by measuring the protein in the column strip using UV absorbance. This method of determining the bound product typically underestimates the amount of product bound during the load due to the gradual desorption of the product during the wash. The product related High Molecular weight (HMW) removal and product recovery results from these experiments are also presented in Figure 13. It is evident from the data presented in Figure 13 that the performance of the cHA step improves significantly with respect to HMW reduction as we move from flow-through conditions to weak partitioning conditions, while product recovery is >80%. Operating under conditions corresponding to a further increase in the partition coefficient (i.e., operating in the binding region) provides no additional benefit with respect to contaminant removal. However, the product recovery across the step begins to drop under strong binding conditions. Thus, the optimum operating window for this separation corresponds to that of weak partitioning chromatography. Under these conditions, a 20-fold reduction of product related HMW species was obtained. The bound product levels under the weak partitioning conditions, in this example, were between 5.1-9.5 mg/ml of the resin. Summary [0206] A second example was presented in hydroxyapatite where operating under weak partitioning chromatography was shown to provide improved performance with respect to HMW reduction with good product recovery (> 80%). Product related HMW species and HMW species bind more tightly to ceramic resin than the product antibody, and is retained strongly under WP conditions. The WP region in this example corresponds to partition coefficients between 8 and 20. [0207] It was once again shown that the performance of the column step can be optimized primarily through the choice of partition coefficients used to run the column. The approach presented here provides a simple means of relating the impact of changing any one of several variables (pH, salt, phosphate, arginine, HEPES etc.,) to column performance. The weak partitioning conditions for optimum performance can easily be identified using the HTS methods described in this example. The approach presented here opens up the possibility of operating in a wider operating space in this mode of chromatography than has been done before. [0208] It is also worth noting here that the concept of weak partitioning chromatography also works in systems that are not driven by charge interactions alone. The 53 WO 2006/099308 PCT/US2006/008919 general approach described in this application can be successfully applied to complex systems such as HIC and hydroxyapatite as well. For example, in addition to the operating pH, several other variables such as NaCl, phosphate salts, arginine / glycine, buffering species, as well as the type of resin can all impact step performance in hydroxyapatite. Nevertheless, one could easily identify the WP window by performing simple batch binding experiments with the product of interest alone. Series 8 ~ Zwitterionic buffer for Protein A elution and subsequent ion exchange steps [0209] A culture containing a monoclonal antibody was purified using MabSelect resin. A Mabselect Protein A column was equilibrated with 5 column volumes of 50 mM Tris/150 mM NaCl, pH 7.5. The column was then loaded at a load of approximately 40 mg product/ml resin. This was followed by a 5CV wash in 1M NaCl, 50mM Tris, pH 7.5 and a 5CV wash containing 10 mM Tris, 75mM NaCl, pH 7.5 wash. The column was then eluted using 30mM HEPES, pH 3.1. The product pool was neutralized to pH 7.2 using 1M HEPES pH 8.0, resulting in a total HEPES concentration of 55mM. At pH 7.2, the HEPES contributes 17mM ionic strength to the buffer. [0210] All references cited herein are incorporated herein by reference in their entirety and for all purposes to the same extent as if each individual publication or patent or patent application was specifically and individually indicated to be incorporated by reference in its entirety for all purposes. To the extent publications and patents or patent applications incorporated by reference contradict the disclosure contained in the specification, the specification is intended to supercede and/or take precedence over any such contradictory material. [0211] All numbers expressing quantities of ingredients, reaction conditions, and so forth used in the specification and claims are to be understood as being modified in all instances by the term "about." Accordingly, unless indicated to the contrary, the numerical parameters set forth in the specification and attached claims are approximations that may vary depending upon the desired properties sought to be obtained by the present invention. At the very least, and not as an attempt to limit the application of the doctrine of equivalents to the scope of the claims, each numerical parameter should be construed in light of the number of significant digits and ordinary rounding approaches. [0212] Many modifications and variations of this invention can be made without departing from its spirit and scope, as will be apparent to those skilled in the art. The specific 54 WO 2006/099308 PCT/US2006/008919 embodiments described herein are offered by way of example only and are not meant to be limiting in any way. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the invention being indicated by the following claims. 55 WE CLAIM: 1. A method of recovering a purified product from a load fluid, comprising the steps of: passing the load fluid through a medium at operating conditions which cause the medium to bind at least 2.8mg of product per mL of medium, wherein the medium is selected from the group consisting of a charged ion exchange medium, a hydrophobic interaction chromatography resin, and an immobilized metal affinity chromatography resin; and recovering the purified product from column effluent. 2. A method of recovering a purified product from a load fluid, comprising the steps of: passing the load fluid through a medium at operating conditions defined by a partition coefficient of at least 0.1; and recovering the purified product from column effluent. 3. The method of claim 1, wherein the operating conditions cause the medium to bind at least 10 mg of product per mL of medium. 4. The method of claim 2, wherein the value of the partition coefficient is in the range of about 0.2 to about 10.0. 5. The method of claim 1 or 2 or any of claims 39 to 44 , further comprising the step of recovering the purified product from the medium using an essentially isocratic wash. 6. The method of claim 1 or 2, or any of claims 39 to 44 wherein the purified product is a protein selected from the group consisting of fusion proteins, Fc-containing proteins, immunoconjugates, cytokines, interleukins, hormones, and therapeutic enzymes. 7. The method of claim 1 or 2, or any of claims 39 to 44 wherein the operating conditions comprise a pH of 7.5-8.2 and a chloride concentration of 12 mM-55 mM. 8. The method of claim 1 or 2, or any of claims 39 to 44 wherein the load fluid further comprises at least one of phosphate, calcium, arginine, glycine, HEPES, and counterligand. 9. The method of claim 8, wherein the counterligand is imidazole. 56 10. The method of claim 1, wherein the operating conditions cause the medium to bind at least 8 mg of product per mL of medium. 11. The method of claim 10, wherein the operating conditions cause the medium to bind at least 14 mg of product per mL of medium. 12. The method of claim 1, wherein the operating conditions cause the medium to bind 2.8 mg of product per mL of medium to 12 mg of product per mL of medium. 13. The method of claim 12, wherein the operating conditions cause the medium to bind 2.8 mg of product per mL of medium to 5 mg of product per mL of medium. 14. The method of claim 12, wherein the operating conditions cause the medium to bind 8 mg of product per mL of medium to 12 mg of product per mL of medium. 15. The method of claim 2, wherein the value of the partition coefficient is in the range of 0.5 to 15. 16. The method of claim 15, wherein the value of the partition coefficient is in the range of 0.5 to 2.5. 17. The method of claim 6, wherein the Fc-containing protein is an antibody. 18. The method of claim 2, wherein the medium comprises a charged ion exchange medium. 19. The method of claim 1 or 18, wherein the charged ion exchange medium comprises an anion exchange resin. 20. The method of claim 1 or 18, wherein the charged ion exchange medium comprises a cation exchange resin. 21. The method of claim 2, wherein the medium comprises a hydrophobic interaction chromatography resin. 22. The method of claim 2, wherein the medium comprises a hydroxyapatite resin. 57 23. The method of claim 2, wherein the medium comprises an immobilized metal affinity chromatography resin. 24. The method of claim 1 or 2, or any of claims 39 to 44 wherein the load fluid comprises one or more impurities selected from the group consisting of host cell proteins, nucleic acids, product variants, endotoxins, Protein A, and viruses, or any combinations thereof. 25. The method of claim 1 or 2, or any of claims 39 to 44 wherein the load fluid is a product-containing fluid eluted from a Protein A column using an elution buffer and the pH and conductivity of the product-containing fluid is adjusted using a neutralization buffer which results in no more than 20mM of the ionic strength of the product-containing fluid. 26 The method of claim 25, wherein the elution buffer comprises molecules with a charged anionic group with a pKa of 2-5. 27. The method of claim 26, wherein the elution buffer further comprises molecules with a charged cationic group with a pKa of 6.5-10. 28. The method of claim 25, wherein the elution buffer comprises a molecule that is a zwitterion at pH 4 to 9. 29. The method of claim 25, wherein the zwitterion is selected from the group consisting of glycine; 1,4-piperazinebis-(ethanesulfonic acid); glycylglycine; cyclopentanetetra-1,2,3,4- carboxylic acid; N,N-bis(2-hydroxyethyl)-2-aminoethanesulfonic acid; 2-(N-morpholino) propane-sulfonic acid; N-tris(hydroxymethyl)methyl-2-aminoethane sulfonic acid; N-2- hydroxyethylpiperazine-N'-2-ethanesulfonic acid; 4-(2-hydroxyethyl)-l-piperazinepropane sulfonic acid; N-tris(hydroxymethyl)methylglycine; glycinamide; N,N-bis(2-hydroxyethyl) glycine; N-tris(hydroxymethyl)methyl-2-aminopropane sulfonic acid; and N-glycyl-glycine. 30. The method of claim 29, wherein the zwitterion is glycine. 31. The method of claim 1 or 2, or any of claims 39 to 44 wherein the product is loaded onto the medium at a concentration of least 100 mg of product per mL of medium. 32. The method of claim 31, wherein the product is loaded onto the medium at a concentration of least 500 mg of product per ml_ of medium. AMENDED PAGE 58 33. The method of claim 32, wherein the product is loaded onto the medium at a concentration of least 1000 mg of product per mL of medium. 34. The method of claim 1 or 2, or any of claims 39 to 44 wherein the concentration of product in the load fluid is at least 1 mg of product per mL of load fluid. 35. The method of claim 34, wherein the concentration of product in the load fluid is at least 10 mg of product per mL of load fluid. 36. The method of claim 35, wherein the concentration of product in the load fluid is at least 100 mg of product per mL of load fluid. 37. The method of claim 1 or 2, or any of claims 39 to 44 wherein at least 89% of the product is recovered from the column effluent. 38. The method of claim 37, wherein at least 95% of the product is recovered from the column effluent. 39. A method of recovering a purified product from a load fluid, comprising the steps of: passing the load fluid through a medium at operating conditions which cause the medium to bind 2.8 mg of product per mL of medium to 13 mg of product per mL of medium; and recovering the purified product from column effluent. 40. A method of recovering a purified product from a load fluid, comprising the steps of: passing the load fluid through a medium at operating conditions which cause the medium to bind at least 1 mg of product per mL of medium, wherein the medium is selected from the group consisting of a hydrophobic interaction chromatography resin and an immobilized metal affinity chromatography resin; and recovering the purified product from column effluent. 59 41. A method of recovering a purified antibody from a load fluid, comprising the steps of: passing the load fluid through a charged anion exchange medium at operating conditions which cause the medium to bind at least 2.8 mg of antibody per mL of medium, wherein the operating conditions comprise a pH of 7.5 to 8.2 and chloride concentration of 12 mM to 55 mM, and wherein the load fluid is a product-containing fluid eluted from a Protein A column; and recovering the purified antibody in the column effluent. 42. A method of recovering a purified antibody from a load fluid, comprising the steps of: passing the load fluid through a charged anion exchange medium at operating conditions defined by a partition coefficient of at least 1.0; wherein the operating conditions comprise a pH of 7.5 to 8.2 and chloride concentration of 12 mM to 55 mM, and wherein the load fluid is a product-containing fluid eluted from a Protein A column; and recovering the purified antibody in the column effluent during the load cycle and any essentially isocratic wash. 43. A method of recovering a purified product from a load fluid, comprising the steps of: passing the load fluid through a medium at operating conditions which cause the medium to bind at least 1 mg of product per mL of medium; and recovering at least 98% of the product from column effluent. 44. A method of recovering a purified product from a load fluid, comprising the steps of: passing the load fluid through a medium at operating conditions which cause the medium to bind at least 3.0 mg of product per mL of medium; and recovering at least 93% of the product from column effluent. 60 This invention relates to methods of using weak partitioning chromatography for the purification of a product from a load fluid containing one or more impurities. Further, the invention relates to methods of weak partitioning chromatography defined by operating conditions which cause a medium to bind least 1 mg of product per mL of medium, or alternatively, defined by a partition coefficient of at least 0.1.