Research Article J Mol Microbiol Biotechnol 2004;7:197–203 DOI: 10.1159/000079828 Effect of Plasmid Copy Number and lac Operator Sequence on Antibiotic-Free Plasmid Selection by Operator-Repressor Titration in Escherichia coli Rocky M. Cranenburgh Kathryn S. Lewis Julian A.J. Hanak Cobra Biomanufacturing Plc, The Science Park, Keele , Staffs., UK Key Words Introduction Recombinant protein · Metabolic burden · DNA vaccine · lac repressor · DAP · DH1 lacdapD The use of antibiotics and antibiotic resistance genes in the production of recombinant biotherapeutics is unde- sirable for a number of reasons. These include the risk of Abstract antibiotic gene transfer to pathogenic organisms [David- The Escherichia coli strain DH1 lacdapD enables plasmid son, 1999], plasmid loss due to plasmid size and metabol- selection and maintenance that is free from antibiotics ic stress [Smith and Bidochka, 1998] or antibiotic break- and selectable marker genes. This is achieved by using down during culture and the risk of contamination of the only the lac operator sequence as a selectable element. product with antibiotics [Baneyx, 1999]. Strategies have This strain is currently used to generate high copy num- been developed to avoid the use of antibiotics, including ber plasmids with no antibiotic resistance genes for use plasmids enabling complementation of a host auxotrophy, as DNA vaccines and for expression of recombinant pro- or possessing post-segregational killing (PSK) mechanisms teins. Until now these have been limited to pUC-based [Hanak and Cranenburgh, 2001]. However, these share plasmids containing a high copy number pMB1-derived another major disadvantage with the antibiotic-selected origin of replication, and the principle lacO1 and auxil- plasmids: the metabolic burden associated with constitu- iary lacO3 operators. In this study we have shown that tively expressed plasmid-borne genes resulting in lower this system can also be used to select and maintain rates of growth [Bentley et al., 1990]. The solution to this pBR322-based plasmids with the lower copy number problem is to use a plasmid with a non-expressed selectable pMB1 origin of replication, and that lacO 1 alone or a pal- marker. The Operator-Repressor Titration (ORT) system indromic version of lacO1 can provide a suffi cient level enables the selection and maintenance of plasmids that are of repressor titration for plasmid selection. This is advan- free from expressed selectable marker genes and require tageous for recombinant protein production, where low only the short, non-expressed lac operator for selection and copy number plasmids are often used and plasmid main- maintenance [Williams et al., 1998]. The principle Esch- tenance is important. The degree of repressor titration erichia coli ORT strain, DH1lacdapD [Cranenburgh et al., due to these plasmids was measured using the natural 2001], has been used to produce several important DNA lactose operon in E. coli DH1 as a model. vaccine candidates such as the HIV-1 vaccine pTHr.HIVA Copyright © 2004 S. Karger AG, Basel [Hanke and McMichael, 2000]. © 2004 S. Karger AG, Basel Rocky M. Cranenburgh 1464–1801/04/0074–0197$21.00/0 Cobra Biomanufacturing Plc Fax +41 61 306 12 34 The Science Park E-Mail [email protected] Accessible online at: Keele ST5 5SP (UK) www.karger.com www.karger.com/mmb Tel. +44 1782 714181, Fax +44 1782 799817, E-Mail [email protected] Although DNA vaccine and gene therapy applications require the highest possible plasmid copy number, recom- binant protein expression can benefi t from using low copy number plasmids instead [Jones et al., 2000], so it was important to determine if these could be selected and maintained by ORT. In addition to plasmid copy num- ber, we investigated the effect of different sequences of lac operators on repressor titration. For those plasmids that do not possess the lacO 1 and lacO 3 operators with the optimum spacing found in the pUC series, we inves- tigated lacO 1 alone and a palindromic version of lacO as potential selectable markers. This would make the con- version of conventionally selected plasmids to ORT plas- mids easier, and reduce the size of the ORT plasmid. The results presented here test the versatility of the ORT sys- tem and determine its potential for recombinant protein production. Fig. 1. The mechanism of repressor titration. A representation of the chromosome of DH1 lacdapD in ( A) the absence, and (B ) pres- ence of a lacO -containing multicopy plasmid. Results and Discussion Plasmid Elements Repressor titration occurs in a cell containing the fol- Virtually all E. coli plasmids in common use as cloning lowing three components: (i) a chromosomal gene under or expression vectors were originally derived from the negative regulation by a promoter that incorporates an natural E. coli plasmid pMB1 [Yoshimori et al., 1972]. operator, (ii) a multicopy plasmid that possesses an op- The pMB1 derivative pBR345 was used to construct erator (with a similar, but not necessarily identical, se- pBR322 [Bolivar et al., 1977]. The pBR322 derivative quence to the chromosomal operator) and (iii) a second pUR1 was the ancestor of the higher copy number pUC chromosomal gene that expresses a repressor protein series [Vieira and Messing 1982], including pUC18 [Nor- that is able to bind to both plasmid and chromosomal rander et al., 1983]. Both the pUC18- and pBR322-de- operators. In DH1lacdapD the chromosomal gene con- rived plasmids constructed for these experiments have a trolled by P lac is dapD [Richaurd et al., 1984], which degree of sequence identity including bla (retained to en- encodes the enzyme tetrahydrodipicolinate N-succinyl able the identifi cation of plasmid-containing cells by an- transferase that catalyses a step in the lysine/diami- tibiotic selection) and the pMB1 origin of replication nopimelate (DAP) biosynthesis pathway. DAP cross- ( ori ). The plasmids were designed to be small and to not links peptidoglycan in the bacterial cell wall, so in the contain any unnecessary expressed elements, so that met- absence of dapD expression, DH1lacdapD cells will lyse. abolic burden would not be a differentiating factor. There- There is no DAP present even in rich media such as fore, the tetracycline resistance gene tet was removed Luria-Bertani (LB) or Terrifi c Broth, as they do not con- from all the pBR322-based plasmids. tain components derived from bacteria. The LacI re- There are two important differences with respect to pressor protein binds to the lacO 1 and lacO 3 operators copy number between the pBR322 and pUC plasmids. and blocks transcription of dapD . Therefore, an un- Twigg and Sherratt [1980] deleted a section of the pMB1- transformed cell of DH1lacdapD can only grow to form derived plasmid pAT153 that resulted in up to a 3-fold a single colony on a nutrient agar plate if an inducer of increase in copy number. This was later found to encode the lac operon such as isopropyl- ß-D-thiogalactopyrani- the Rom protein, present in pBR322 but absent from side (IPTG) is present. However, when a plasmid pos- pUC18. The pMB1 ori encodes two overlapping RNA sessing lacO sequences is introduced into the cell, the transcripts: RNAII is the primer for plasmid DNA syn- binding of the repressor protein to the plasmid-borne thesis, and RNAI is the antisense inhibitor of RNAII [To- operator derepresses the chromosomal operator and al- mizawa, 1984]. Rom stabilises the binding of RNAI to lows dapD expression ( fi g. 1 ). RNAII [Tomizawa and Som, 1984]; in the absence of 198 J Mol Microbiol Biotechnol 2004;7:197–203 Cranenburgh/Lewis/Hanak 5,000 LB 4,500 Ap +IPTG 4,000 3,500 3,000 2,500 Fig. 2. Transformation effi ciency of 2,000 DH1lacdapD . Following transformation, 1,500 cultures were plated onto LB agar with no 1,000 (colonies per fmol DNA) 500 additives (ORT selection) and with ampi- Transformation efficiency 0 cillin (Ap) plus IPTG (antibiotic selection). I I t 8 8 1 0 0 2 1 e 1 1 0 c c 2 0 t C C 8 la la 3 2 ∆ Results represent the number transformant U 1 ∆ 2 R 2 2 p U C 8 2 3 2 p U 1 3 B R 3 p C R p B R colonies per femtomole of plasmid DNA U B p B used in the transformation, and are the p p p mean of three separate experiments with Plasmid standard deviation displayed as error bars. Rom, RNAII is able to dissociate from RNAI, and there- Table 1. Plasmids used in this study fore to prime plasmid replication, at a greater rate. The second factor is a G KA mutation in the pMB1 ori from Plasmid Size, bp Operator Source or reference pBR322 to the pUC series [Minton et al., 1988] that pUC18 2,686 lacO , lacO Norrander et al., 1983 - 1 3 causes a further copy number increase in rom plasmids pUC18O1 2,085 lacO1 This study by altering RNAII conformation [Lin-Chao et al., pUC18I 2,084 Ideal lacO This study 1992]. pUC18¢lacO 2,364 None Cranenburgh et al., 2001 pBR322lacO 3,188 lacO1, lacO3 This study These mutations enable pUC plasmids to reach 500– pBR322O1 3,104 lacO1 This study 700 copies per cell [Minton et al., 1988]. The single chro- pBR322I 3,103 Ideal lacO This study mosomal copy of lacI expresses LacI repressor protein at pBR322¢tet 2,866 None This study ~20 copies per cell [Müller-Hill et al., 1968]. It was previ- ously thought that this signifi cant molar excess of plasmid was necessary to achieve suffi cient levels of repressor ti- tration in ORT E. coli strains. It was not known if lower with the optimal spacing (71 bp between the elements), copy number pBR322 plasmids with 39–55 copies per (ii) a single lacO 1 , (iii) a single ‘ideal’ lacO , and (iv) no cell [Lin- Chao and Bremer, 1986] would achieve the same lacO sequences, as described in table 1 .