Gene Knockouts and Exchanges by Linear Transformation

6/24/2003 ECK
revised 4/7/2008


Genes can be altered by recombination with linear DNA molecules. This requires a high internal DNA concentration, achievable by electroporation. The lambda red system allows efficient recombination between homologous sequences as short as 40 bp, which frees us of the need to provide long tracts of homology for recombination into the chromosome.

By using conditions favoring efficient electroporation of appropriately designed PCR products into lambda red backgrounds, we can manipulate virtually any gene of choice directly in the chromosome and without requiring a selectable phenotype. All we need to know ahead of time is the sequence of the gene and its immediate neighborhood.

The process is stepwise extensible and iterative. You can make very complex constructions. Intermediates can be rapidly cloned into TOPO plasmids, sequenced for fidelity and modified again by linear transformation. Products can be easily introduced into the chromosome by the same technique.


1. Preparation of linear DNA

The example given is construction of a linear DNA molecule for simple replacement of a gene A by a drug cassette R. The design details are illustrated in the figure.

Pick two regions, about 40 bp each, between which you wish to introduce the cassette. Call these "a1" and "a2". Design primers to amplify a drug cassette R. It is best if this cassette is already in a bacterial chromosome, as this minimizes the risk of phage or plasmid contamination during the subsequent electroporation. If it is necessary to use a phage, plasmid or similar autonomous element as cassette source, band purify the product and then reamplify it.

Add tract a1 as a 5' tail to the upstream drug primer, yielding primer P1. Add the complement of a2 as a 5' tail to the downstream drug primer, yielding P2. Use P1 and P2 to amplify the cassette. The product will have tails at each end providing homology blocks for red-mediated recombination directly into recipient DNA, which can be either the chromosome or another plasmid.

Universal Cassettes & Primers. Designing cassette-specific primer pairs for each knockout can be expensive. We have made a set of "universal" cassettes by appending standardized caps to drug resistance genes embedded in a common regulatory context where possible, the exception being tetAR. See "Drug Cassettes" for more details. These cassettes do not recombine with Salmonella or E.coli chromosomes. They are amplified using primers with cognate universal 3' ends and 5' tails providing the required transformation homology blocks. Primers which work for one cassette will work for all. In addition, primers can be shared by different research groups, as long as they are designed with these standardized ends.

The caps we use are:



They have been chosen to minimize internal secondary structure, to avoid low-stringency "wobble" pairing with false sites, and to prevent cross-reaction with each other leading to primer dimers.

Once one cassette is made with these termini, it can be reamplified with appropriate primers having the same 3' tracts. It can also be converted to another drug type by a precise linear transformation of the coding region. Starting with the chloramphenicol resistance cassette described earlier, we made a "universal" version, and then converted it to resistance to kanamycin, zeomycin, streptomycin, rifampicin and gentamycin, all with the same promoter and associated upstream and downstream sequences. This provides us with a half dozen drug resistance cassettes all with the same constitutive regulatory sequences of proven efficiency and all of which can be amplified by any pair of "universal" primers.

When a "universal" drug cassette is finally inserted into the chromosome, we have found that the chromosomal DNA itself becomes an excellent template for future amplifications, as it has survived selection for drug resistance and is absolutely uncontaminated by plasmids or primers left over from previous amplifications. In a pinch, you can simply poke the ice of a frozen strain as a source of quick template.

Examples of universal primers used to knockout dinB of Salmonella and replace it with any universal-tailed drug cassette would be:



The homology blocks for both primers were designed in a manner exactly analogous to the two primers described earlier. The first 40 bp of TP1297 are the first 40 bp of the dinB gene, and the first 40 bp of TP1298 are the reverse-complement of the last 40 bp of the same gene.

Problems with second-drug transformations. Because most universal cassettes share considerable sequence identity, a second transformation into a background already containing such a cassette can be problematic, as the new drug will frequently swap out the first one, rather than insert into its desired target. There are several workarounds. Perhaps the easiest is to maintain selection on the first drug. Drawbacks are decreased efficiency and occasional unintended selection of duplications. Another method is to use the tetRA cassette as one of the two drugs, with the obvious limitations of larger cassette size, need for induction, and toxicity of overexpressed TetA. Finally, a specialized drug cassette can be designed which has the native control region associated with the drug resistance gene, but is still terminated by universal ends. We have done this with good success for drugs such as arr-2 (rifR). We have not found the caps themselves to be a problem in linear transformation. Their size makes them inefficient substrates for the red system.

Hybridization stabilization trick. Pat Higgins points out that the interaction of a 40 bp homology block with its target site can be stabilized if the hybrid contains a C:C mismatch or GATC tract. In both cases, the stabilization is the result of a host DNA binding protein adhering nonproductively to the region, although the identity of the protein is known only in the second case, in which Dam methylase binds tightly to the hemimethylated hybrid.

Don't use phage or plasmid DNA directly as cassette template! A few of the original molecules will always make it through the amplification process, giving false positives after electroporation. Instead, make a batch of cassette DNA either by amplification with short primers (e.g., camL and camR for the chloramphenicol cassette of pACYC184), or by cutting the plasmid to create a template fragment lacking the plasmid origin of replication. Band purify the product following agarose electrophoresis. I usually soak the excised band overnight in an equal volume of TE. This can be stored at -20° and used in a 1:100 dilution as template for amplification of tailed cassettes. For instance, TP557 and TP558 yield a 1.4 KB product, comprised of a chloramphenicol resistance cassette flanked by 40 bp blocks of homology to the respective ends of the hisG gene of Salmonella.

Eliminate primers prior to electroporation! They can interfer with the transformation. Generally, a clean up with a Qiaquick or Wizard PCR purification kit is sufficient. If primers are clearly contaminating the final preparation, then add to every 50 µL of product 5 µL of 10x PCR buffer and 1 µL of Exonuclease I (USB/Amersham), diluted to 1 u/µ. Incubate 1 hr at 37°C, and clean the product again with a PCR purification kit. The digest can be done directly in the PCR reaction mix following amplification, if desired.

2. Lambda red-expressing cells

Plasmid pKD461 expresses lambda red under Para control. Expression is regulated by L-arabinose. The plasmid carries bla and can be maintained under ampicillin selection. The plasmid origin is temperature sensitive, and the plasmid will be lost on temperature shift from 30°C to 42°C. Other versions exist with different antibiotic selections.

The following pKD46-carrying strain is fully r-m+, which allows it to tolerate unmethylated or inappropriately methylated foreign DNA. It also has a higher electroporation efficiency than LT2 for unknown reasons:

TT22971 sty(LT2) metA22 metE551 trpD2 ilv-452 leu- pro-(leaky) hsdLT6 hsdSA29 hsdB strA120 $pKD46 araC bla oriR101 repA101ts lambda red (gam+ bet+ exo+) $COM grow at 30°C!; lambda red under Para control.

3. Electroporation-competent cells

Grow a fresh overnight of TT22971 in 2 mL LB + ampicillin@50 µg/mL at 30°C.

Dilute 0.5 mL of overnight into 50 mL LB + ampicillin@50 µg/mL and 10 mM L-arabinose.

Shake 3 hr at 30°C in ribbed Erlenmeyers with good head space.

Wash 3x in 20 mL ice-cold 10% glycerol made in stringently deionized water (with resulting low conductivity). Do not resuspend the last pellet. Instead, vortex it vigorously until it forms a muddy slurry. Adjust the volume to accomodate the number of transformations required.

The cells lose their efficiency when frozen, and should be used within a few hours.

4. Electroporation

Add 2 µL of linear DNA to 40 µL cells and transfer to an ice-cold electroporation cuvette.

Follow the standard electroporation protocol.

Let the cells recover 2-4 hours at 37 °C; in 1 mL LB. Do not include ampicillin or L-arabinose! From here on, we wish to get rid of red.

Pellet the cells and resuspend in 100 µL 10% glycerol.

Plate to selective medium and incubate at 42°C.

Streak selected colonies for singles, test for loss of pKD46 by ampicillin sensitivity, and screen by PCR using appropriate primers.


i) Controls

-- Electroporation: Add phage DNA containing the cassette you are using directly to an aliquot of competent cells, and test how efficiently the cells are transformed to drug resistance. Always include this control. The efficiency of electroporation is the single most important variable in linear transformation.

-- Transformation: Replace hisG with the chloramphenicol cassette described above. Select chloramphenicol resistance, and score for his- by printing to minimal/glucose and then to minimal/glucose + his. This particular knockout has been done repeatedly and is well-characterized. Do it whenever you are having trouble with a transformation. If it works, then there is something unusual about the gene you are trying to eliminate. For instance, it may be an essential gene, or the homology blocks being used may be duplicated elsewhere in the genome. If the control doesn't work, then you've made a mistake somewhere in the protocol.

ii) red is probably mutagenic in the long run, as it reshuffles DNA through short homologies, creating random garbage. You should move your modified gene into a clean background by P22 transduction.


This technique borrows heavily from methods developed for linear transformation of E.coli, several of which use red to facilitate recombination between short homologies1,2,3,4,5.


1. Datsenko, K.A. and Wanner, B.L. (2000). One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc. Natl. Acad. Sci. 97, 6640-6645.

2. El Karoui, M. et al. (1999). Gene replacement with linear DNA in electroporated wild-type Escherichia coli. Nuc.Acids Res. 27, 1296-1299.

3. Murphy, K.C. (1998). Use of bacteriophage lambda recombination functions to promote gene replacement in Escherichia coli. J. Bacteriol. 180, 2063-2071.

4. Poteete, A.R. and Fenton, A.C. (2000). Genetic requirements of phage lambda red mediated gene replacement in Escherichia coli K-12. J. Bacteriol. 182, 2336-2340.

5. Poteete, A.R. and Fenton, A.C. (1984). Lambda red-dependent growth and recombination of phage P22. Virology 134, 161-167.