Lentiviral expression clones, which contain long direct repeats, often show dramatic instability in Escherichia coli, leading to difficulties in obtaining valid clones. We show that the reduced-genome E. coli strain MDS42 is capable of stabilizing lentiviral expression clones containing direct repeats, and outperforms many commonly used cloning strains for this purpose. In addition, the strain has several characteristics that make it highly amenable for use in recombinational cloning systems.

Lentiviral expression vectors generally contain large regions of directly repeated DNA sequence from the long terminal repeats (LTRs) of the retroviruses from which they are derived (1,2). Cloning of these DNA sequences in standard Escherichia coli cloning hosts (such as DH5α or TOP10) leads to plasmid instability because of deletion of the regions between the LTRs, presumably due to homologous recombination events (3). Since such deletion removes most of the important lentiviral sequences and produces a smaller plasmid usually containing only the antibiotic resistance marker and replication origin, the recombinant plasmids are strongly selected, and recovery of the desired clones is difficult. While strains of E. coli, such as STBL3 (Invitrogen, Carlsbad, CA, USA) have been developed to overcome some of these problems, these strains are often slow growing, sensitive to phage T1, difficult to make transformation competent, or fail to solve the recombination problem completely (4) (also see www.invitrogen.com/content/Focus/Focus%20Volume%2016%20Issue%203.pdf). Recently, a reduced-genome strain of E. coli called MDS42 was developed that claimed to reduce plasmid recombination events, possibly due to the loss of insertion sequence (IS) elements, which might be causing higher levels of recombination (5). The data shown dealt mainly with inverted repeat sequences, and we sought to identify whether MDS42 could enhance stability of direct repeat sequences and if it was amenable to use as a cloning strain for Gateway® recombinational cloning.

E. coli MDS42 and MDS42recA were purchased from Scarab Genomics LLC (Madison, WI, USA), and chemically competent cells were generated by the procedure of Hanahan et al. (6). We were able to generate competent cells with a transformation efficiency of 1 × 10⁸ cfu/µg, which is two to three times better than the transformation efficiencies we can obtain with STBL3 cells using this procedure. MDS42 and MDS42recA also grew faster at both 30° and 37°C than STBL3, with doubling times for MDS42recA being 40 and 30 min, respectively, compared with 55 and 40 min for STBL3. Both reduced-genome strains, as well as STBL3, TOP10, and DH5α were used to test for plasmid stability with a lentiviral expression clone called pEL101–693. This vector (see Figure 1) contains an enhanced green fluorescent protein (eGFP) reporter gene in a backbone derived from pLenti4-BlockIT-DEST (Invitrogen), which contains 183- and 109-bp direct repeats from the U3 and U5 regions of the viral LTRs (7). A Gateway LR recombination reaction (Invitrogen) was carried out to generate the lentiviral expression clone; 1 µL LR reaction was transformed into the cell lines, and the sample was plated on LB-agar plates containing 100 µg/mL ampicillin for selection. Plates were incubated for 24 h at 30°C.

Figure 1. Diagrams of the expected 7.4-kB expression clone (pEL101-693) and the recombinant 3.6-kB expression clone created by LTR recombination (Click to enlarge)

All plates contained colonies after 24 h: TOP10, DH5α, MDS42, and MDS42recA colonies had similar morphologies, while STBL3 colonies appeared to have two distinct morphologies—a round tan colony similar to normal E. coli colonies, and a flat white colony which we often observe using STBL3 cells. In general, we have observed that the flat white colonies are indicative of the desired clones, while the round colonies contain plasmids that have recombined out the lentiviral DNA. Several colonies were picked from each plate and were grown at either 30° or 37°C in Superior Broth™ (AthenaES, Baltimore, MD, USA) overnight. Plasmid DNA was prepared using the FastPlasmid® kit (Eppendorf, Hamburg, Germany), and Table 1 shows the results of agarose gel electrophoresis analysis of the colonies. In some cases, a plasmid of the expected 7.4-kB size was observed. In other cases, a recombinant 3.6-kB plasmid was the only DNA seen, or a mixture of 7.4- and 3.6-kB plasmids was observed. Finally, in some cases with STBL3, we saw no DNA on the gel—it is unclear whether this represents a dramatic decrease in DNA copy number, or possibly loss of the plasmid. Recombination of the long direct repeat in this expression clone would produce a 3.6-kB plasmid containing the ampicillin resistance gene and bacterial origin of replication (see Figure 1). Restriction analysis of the plasmids confirmed that the 3.6-kB plasmids represented recombination events at the 183-bp LTR direct repeats, while the 7.4-kB plasmids represented the desired full-length expression clones. Several of the 7.4-kB plasmids were also sequenced through the LTR regions and the eGFP reporter region, and contained the expected DNA sequences with no evidence of recombination. Two of the 3.6-kB plasmids were sequenced, and this confirmed that they did in fact have a recombinant LTR—in one case, the plasmid had lost nearly all of the LTR sequences, and in the other, a 75-bp duplication of part of the LTR repeat was present.

Table 1. Plasmid Sizes of Lentiviral Clones Produced in Different Escherichia coli Strains at Indicated Temperatures (Click to enlarge)

MDS42 clearly outperforms DH5α and TOP10 and is equivalent or better than STBL3 at both 30° and 37°C. To further these findings, we carried out a multisite Gateway recombination reaction to introduce both a promoter and a reporter gene into a lentiviral vector with a different backbone and set of repeat elements, derived from the pSico vectors (8) and containing a 189-bp direct repeat. Multisite Gateway recombination requires higher competence E. coli and yields fewer colonies per reaction, making LTR recombination more of an undesirable problem. A reaction was transformed into both MDS42recA and STBL3 cells, and both were plated on LB-agar containing 100 µg/mL ampicillin. Table 2 shows the results of plasmid preparations of 12 colonies from four different reactions [each with a different combination of cytomegalovirus (CMV) or Rosa promoter and eGFP or monomeric red fluorescent protein (mRFP) reporter]. In all cases, MDS42 outperformed STBL3 in producing colonies containing the expected size DNA and also had much higher rates of transformation. Results were similar with the RecA⁺ and recA MDS42 strains which was unexpected, as the use of recA strains for cloning has always been viewed as a necessity for avoidance of direct repeat recombination. This argues strongly for the claim of Posfai et al. that other IS elements may be responsible for most of this activity in E. coli (5). Nevertheless, we have chosen to use the MDS42recA strain as our strain of choice for cloning repeat-containing plasmids. Previously, using STBL3 cells, we often had to pick four to six clones in order to ensure an accurate expression clone was obtained. We have since carried out more than 100 cloning reactions using MDS42 in which two colonies have been chosen for analysis—in >95% of the cases both clones chosen were the expected size, and many of these clones were used to generate functional lentiviruses, saving a significant amount of effort in plasmid preparation.

Table 2. Plasmid Sizes of Multisite Gateway Lentiviral Expression Clones Produced in Different Escherichia coli strains at 30°C (Click to enlarge)

Taken together, these data argue for the utility of the MDS42 reduced-genome strains in stabilizing lentiviral expression constructs. MDS42 strains have some of the advantages of DH5α, such as rapid doubling time, T1 resistance, and high chemical competence, while duplicating or improving on the DNA stability aspects of strains such as STBL3. Most importantly, these strains minimize the number of DNA constructs that need to be prepared for a given clone, which can dramatically improve efficiency of even moderately high-throughput cloning.