Downstream purification

1.3.2.1 Intermediate purification

The removal of major contaminants, following initial plasmid release (as described in Section 1.3.1.2), may be effected in various ways. As stated above, the alkaline lysis method of Birmboim and Doly (Birnboim and Doly 1979; Birnboim 1983) is almost universally used as the initial cellular disruption step and has the advantage that a large proportion of cellular debris, proteins, and large chromosomal DNA fragments are precipitated. The resulting insoluble floe produced at this stage may be readily removed by centrifugation (QIAGEN 07/1999), or flotation and filtration (Theodossiou et al. 1997; Varley et al. 1999).

The remaining contaminants at this point in the process can be removed in a variety of ways. RNA is a major contaminant, and is not completely eliminated by precipitation as a result of the alkaline lysis procedure. At laboratory scale digestion with RNAse is commonly used to break the RNA polymer into small monomers, which are more readily separated (QIAGEN 07/1999; Sambrook et al. 1989). Enzymatic digestion is however likely to be expensive at industrial scale. Also, the bovine pancreas is a common source of RNAse. The use of enzymes derived from such sources will not be acceptable to the regulatory authorities in the production of pharmaceutical grade products due to potential risk with regards to the presence of mammalian pathogens such as the prion which causes BSE (Durland and Eastman 1998). In order to avoid addition of exogenous RNAse Monterio and co-workers (Monterio and al 1999) showed that the activity of endogenous nucleases remaining following the removal of the floe formed in the alkaline lysis step can result in up to a 40 % w/w reduction in host RNA levels when lysates are incubated at 37°C. This however resulted in 9 % w/w plasmid loss. More recently the

modification of an E. coli host strain to include an expression cassette for

al. 2001). The expressed RNA is translocated to the periplasm of the cell from whence it is released upon lysis and acts to hydrolyse the bulk of host RNA. This methodology avoids the use of exogenous animal derived RNAse and thus should be acceptable to the regulatory authorities.

There are several alternative methods for the removal of RNA that do not rely on endogenous RNAse activity. The ability of diatomaceous earth to absorb RNA in favour of SC plasmid DNA has been the basis of a recent patent (Horn et al. 1996). The retention of plasmid DNA while low molecular weight contaminants are removed with the permeate during ultrafiltration has also received considerable investigation. Bussey and co-workers (Bussey 1998) described the use of tangential flow filtration, using membranes with typical molecular weight cut off (MWOO) between 300-500 kDa, for the purification of plasmids up to 50 kb. Contaminants such as proteins and low molecular weight RNA pass through the membrane while the plasmid is retained. However shear may be a problem, particularly with larger plasmids. Kahn and co-workers (Kahn et al. 2000) have recently combined a filtration process with extended exposure to alkaline conditions during the alkaline lysis step (with the effect that remaining RNA contaminants were reduced in both size and abundance). Tangential flow filtration was carried out with a polyethersulfone membrane with nominal MWCO between 500 and 1000 kDa (for plasmids in the range of 5 . 6 - 1 0 kb). 0.5 ft^ of membrane was required per 10-15 g cells (7-20 mg plasmid) processed. Greater then 99 % w/w RNA and 95 % w/w protein was removed; however, endotoxin levels remained high (2400 ± 1700 EL) mL‘^) and no mention is made of the reduction in the levels of chromosomal DNA contamination, which appear to be high from the analytical gels published. The initial depletion of RNA levels in the process liquor before the high-resolution stages may, however, prove useful.

Purification of plasmid DNA on the basis of differential solubility's compared to contaminant molecules has been exploited through fractional precipitation using Isopropyl alcohol (IPA), various salts (such as lithium chloride.

ammonium acetate or ammonium sulphate), polymers such as polyethylene glycol (PEG) or combinations of these (Durland and Eastman 1998; Ferreira et al. 2000; Levy et al. 2000c). The use of the compaction agents spermidine and spermine (small cationic molecules which bind to the minor grooves of a dsDNA molecule, resulting in 4 - 6 orders of magnitude reduction of the volume occupied by the DNA molecule) has also been described (Murphy 1999). Shear, resulting in plasmid losses during resuspension of precipitated plasmid is again a danger at industrial scale. Work carried out by Collins and co-workers (Collins et al.) utilised a two step PEG precipitation process to further purify clarified lysates containing a 6.9 kb plasmid. 5% w/v PEG precipitation first removed contaminating DNA and protein, while the subsequent 8 % w/v PEG step selectively precipitated the bulk of the plasmid from remaining contaminants. The method has the advantage that at this point the plasmid precipitate can be resuspended in a buffer suitable for the next unit operation. The removal of contaminants in this way was found to decrease the competition for binding sites on a chromatographic matrix during subsequent purification increasing the capacity of the matrix for plasmid by a factor of 2 (Collins et al.). More recently, McHugh and co­ workers (McHugh and Hoare 2001) investigated the use of CaCb for the selective precipitation of contaminants present in plasmid containing process liquors, while the plasmid remains the supernatant. Addition of 0.2M CaCb to a process stream, which had previously been clarified and concentrated by I PA precipitation, resulted in a purification factor of approximately 6.5. This was mainly due to the precipitation of RNA, with some clearance of single­ stranded chromosomal DNA also being observed (McHugh, personal communication. 2000; McHugh and Hoare 2001). The fractionation of DNA has also been achieved by aqueous two-phase extraction, either using systems comprised of the immiscible polymers PEG and dextran (Rudin and Albertsson 1966; Favre and Pettijohn 1967; Walter et al. 1985), or PEG-salt systems (Cole 1991; Andrews et al. 2001). These results will be further discussed in Section 1.4.2.

1.3.2.2 High-resolution purification

High-resolution techniques for the separation of the plasmid product from chromosomal DNA and other contaminants, such as protein and endotoxin can be achieved through the use of chromatography. In some cases resolution of OC and SC plasmid has also been achieved. The use of anion

exchange chromatography (Ferreria et al. 2000b), size exclusion

chromatography (Horn et al. 1995; Ferreira et al. 1997) and hydrophobic interaction chromatography (Diogo et al. 1999), and also reverse-phase HPLC (Green et al. 1997) have been described for the purification of pharmaceutical grade plasmid DNA. Less conventional methods, such as expanded bed anion exchange chromatography or fluidised bed adsorption (Varley et al. 1999; Ferreria et al. 2000a; Theodossiou et al. 2000; Thwaites et al. 2001), membrane chromatography systems (Van-Huynh et al. 1993; Giovannini et al. 1998; Nochumson et al. 2000), magnetic beads (Levison et al. 1998), triple helix affinity chromatography (Wils et al. 1997; Simon et al. 2001), affinity chromatography utilising a zinc finger-glutathione S- transferase fusion protein as the ligand (Woodgate et al. 2002) and the combination of conventional chromatography mass transfer effects in the presence of an electric field (Cole et al. 2000; Park 2001 ) have also been described.

Anion exchange chromatography is efficient at separating plasmid products from other biological molecules such as proteins, due to the much higher charge typically carried by nucleic acids. Small RNA molecules are also well resolved. However this method is less efficient with regard to large RNA and chromosomal DNA fragments of similar size to the plasmid which may have a similar charge. Endotoxins may also be difficult to remove by this method (Durland and Eastman 1998). Similarly, size exclusion chromatography will not adequately resolve contaminants of similar size to the plasmid product (Durland and Eastman 1998). Reverse phase high performance liquid chromatography can be an extremely effective purification step, but utilises

reagents which are both toxic and expensive, and the requirement for high pressure will complicate scale up, and may result in shear damage to large plasmids. Hydrophobic interaction chromatography has been shown to be efficient at separating plasmid that has been denatured, for example by localised pH extremes during lysis (to the single stranded form) (Diogo et al. 1999). Diogo and co-workers attest that ongoing work indicates the utility of the operation to separate RNA and chromosomal DNA from plasmids. Triple helix affinity chromatography (Wils et al. 1997; Simon et al. 2001) is efficient at purifying plasmid product from RNA, chromosomal DNA and endotoxins, but the matrix is costly, and the technique requires that the plasmid be engineered to contain a homopurine sequence for the affinity separation. No differentiation between plasmid forms is possible using this method. The use of various chromatographic steps in whole plasmid purification processes is considered in more detail in the following section.